Cardiologists

29-1212.00
Median wage $496,010/yr17,290 employed (US)Rank #655 of 923 scored · top 71% by substitution

Diagnose, treat, manage, and prevent diseases or conditions of the cardiovascular system. May further subspecialize in interventional procedures (e.g., balloon angioplasty and stent placement), echocardiography, or electrophysiology.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution21
Exposure22
Augmentation64

Substitution — the headline: capability discounted by cost, barriers and adoption.

Exposure — technical capability alone, regardless of whether anyone deploys it.

Augmentation — how much AI assists without replacing. High here + moderate substitution = a changing job, not a disappearing one.

Tasks on the substitution scale

25 rated tasks, binned by substitution score.

Position among all scored occupations

Distribution of 923 occupation scores; the marker is this occupation.

Tasks with substitution ≥ 70

4%

Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.

Why this score

The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.

Task automatabilityw 35%20

panel mean rating 1.8/5 → substitution pressure 20/100

Technical feasibility todayw 20%26

panel mean rating 2.0/5 → substitution pressure 26/100

Cost vs. human wagew 15%22

panel mean rating 1.9/5 → substitution pressure 22/100

Adoption barriersw 20%inverted — strong barriers lower the score13

panel mean rating 4.5/5 (barrier strength) → substitution pressure 13/100

Sector adoption velocityw 10%27

panel mean rating 2.1/5 → substitution pressure 27/100

Task breakdown (25 tasks)

Substitution pressure per task, weighted by O*NET importance in the composite. Expand a task for the full rater audit trail — every rating, every model, every rationale.

Calculate valve areas from blood flow velocity measurements.

77

CI 74–81 · exposure 80 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Cardiac imaging centers and large cardiology practices have rapidly adopted automated valve measurement in echocardiography platforms over the past decade; this is now standard in most digitized cardiology settings in developed healthcare systems.
Sector adoption velocityclaude-sonnet-55/5Automated valve area calculation is already standard, deeply embedded practice across cardiology and echo labs worldwide.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments cardiologist productivity by eliminating manual measurement and calculation time while surfacing accurate baseline values for clinical interpretation, allowing the physician to focus on clinical context and decision-making rather than computation.
Augmentation potentialclaude-sonnet-55/5Automated calculation significantly speeds up the cardiologist's workflow, letting them focus on image acquisition and clinical interpretation rather than manual math.
Task automatabilityclaude-haiku-4-5-202510014/5AI can reliably perform the computational conversion of blood flow velocity to valve area using established Doppler echocardiography formulas (continuity equation, pressure half-time method). This is a well-defined mathematical task with standardized inputs and outputs, though interpretation of borderline or noisy measurements may require human review.
Task automatabilityclaude-sonnet-54/5This is a deterministic calculation (e.g., continuity equation from Doppler echo data) that software already performs automatically once measurements are entered, meeting the time-saving bar for the calculation step itself.:
Adoption barriersclaude-haiku-4-5-202510013/5Clinical and regulatory expectations require a licensed cardiologist to review and validate measurements before clinical decision-making, and liability concerns favor human sign-off on critical hemodynamic calculations, creating material adoption friction despite technical capability.
Adoption barriersclaude-sonnet-53/5While the calculation itself is automated, a licensed cardiologist must still verify measurement inputs and interpret results for diagnosis and billing, creating moderate oversight requirements.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI-based calculation has negligible marginal cost per measurement once integrated into imaging systems, whereas a cardiologist's time for manual measurement and calculation costs significantly more; the cost difference is orders of magnitude in favor of AI.
Cost vs. human wageclaude-sonnet-55/5The calculation is embedded in existing imaging software at near-zero marginal cost compared to a physician manually computing it.
Technical feasibility todayclaude-haiku-4-5-202510014/5Multiple deployed cardiac analysis software platforms (GE Healthcare, Philips, Canon) incorporate automated valve area calculations from echocardiography data in clinical workflows. These systems are in production at scale, though cardiologist verification of measurements remains standard practice.
Technical feasibility todayclaude-sonnet-55/5Echocardiography machines and analysis software (e.g., from GE, Philips, Siemens) routinely auto-calculate valve areas from Doppler velocity data in clinical production today.

Obtain and record patient information, including patient identification, medical history, and examination results.

56

CI 50–61 · exposure 55 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Large hospital systems and cardiac clinics have rapidly adopted AI-assisted documentation and EHR automation over the past 3–5 years, with measurable displacement of clerical tasks and widespread pilot programs in production.
Sector adoption velocityclaude-sonnet-53/5Healthcare has been a slower-adopting sector overall, but ambient documentation tools have seen notable uptake in outpatient and specialty practices including cardiology in the last two years.
Augmentation potentialclaude-haiku-4-5-202510014/5AI substantially augments cardiologists' productivity through voice-to-text, automated field population from previous records, and decision-support prompts for missing data, keeping the physician in control while reducing manual documentation burden.
Augmentation potentialclaude-sonnet-55/5AI scribes substantially reduce documentation burden and improve note quality/completeness while the physician remains responsible for verifying and finalizing the record.
Task automatabilityclaude-haiku-4-5-202510013/5AI can automate significant portions of data entry and structured record organization from patient intake forms, prior records, and basic examination data, but requires human validation and typically cannot independently conduct the full patient interview or physical examination that generates this information.
Task automatabilityclaude-sonnet-53/5AI scribes and ambient documentation tools can transcribe and structure patient history and exam findings, but obtaining information (interview, physical exam) still requires clinician presence and judgment.
Adoption barriersclaude-haiku-4-5-202510013/5HIPAA compliance, clinical liability for data accuracy, and institutional preference for human-verified records create moderate friction; however, no strict legal requirement mandates human performance of data entry itself, only data governance oversight.
Adoption barriersclaude-sonnet-53/5Documentation must be attributable to and verified by a licensed physician for medical-legal and billing purposes, creating oversight requirements though not an outright prohibition on AI assistance.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-powered documentation and data entry systems cost substantially less than human clerical or nursing time spent on record-keeping, making the ratio strongly favorable to automation when integration costs are amortized.
Cost vs. human wageclaude-sonnet-53/5Ambient scribe subscriptions are cheaper than a human scribe but add licensing costs and still require physician review time, keeping savings moderate rather than order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature EHR systems with AI-assisted documentation, voice transcription, and structured data extraction are widely deployed in cardiology practices and hospitals, though they require human review and occasional correction of errors in complex cases.
Technical feasibility todayclaude-sonnet-53/5Ambient clinical documentation products (e.g., DAX, Nuance) are deployed in real practices and reliably generate notes from visits, though accuracy checks and edits are still routine.

Compare measurements of heart wall thickness and chamber sizes to standards to identify abnormalities, using the results of an echocardiogram.

52

CI 36–67 · exposure 55 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Major cardiac imaging centers and some hospital networks have begun piloting and deploying AI measurement tools, but uptake remains uneven; many smaller labs and independent practices still rely on manual measurement, indicating middling real-world adoption velocity.
Sector adoption velocityclaude-sonnet-53/5Healthcare imaging is adopting AI-assisted measurement and analysis tools at a moderate pace, with growing but not yet ubiquitous integration into echocardiography workflows.
Augmentation potentialclaude-haiku-4-5-202510015/5AI-assisted measurement dramatically raises cardiologist productivity by automating tedious manual tracings and providing instant automated measurements and flagging, allowing the physician to focus on clinical interpretation and borderline cases—a textbook augmentation scenario.
Augmentation potentialclaude-sonnet-54/5AI measurement tools significantly speed up and standardize wall thickness and chamber size quantification, letting cardiologists focus on interpretation and clinical correlation while the human remains the final decision-maker.
Task automatabilityclaude-haiku-4-5-202510014/5AI systems can reliably measure cardiac chamber dimensions and wall thickness from echocardiograms and flag deviations from normal ranges with high accuracy, saving significant time on the measurement and initial screening phases. However, clinical interpretation of borderline cases and integration with patient history typically still requires cardiologist judgment, preventing a full 5 rating.
Task automatabilityclaude-sonnet-52/5AI can measure and flag abnormal values from echocardiogram images with reasonable accuracy, but the diagnostic synthesis with clinical context and final interpretation still requires physician judgment, so full end-to-end automation at equal quality is not yet achieved.
Adoption barriersclaude-haiku-4-5-202510013/5While echocardiograms require a licensed sonographer to acquire, the measurement and initial abnormality detection task itself is not legally restricted to a cardiologist by regulation, though professional liability and clinical credentialing norms create moderate friction in adoption.
Adoption barriersclaude-sonnet-54/5Diagnostic interpretation of echocardiograms is a licensed medical act with significant liability; a cardiologist must review and sign off on findings, creating strong regulatory and liability barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI inference and integration costs are typically $5–50 per study, while a cardiologist's time to manually measure and interpret is $50–150 in loaded labor cost; AI is substantially cheaper but integration and QA overhead prevent a full 5-fold margin.
Cost vs. human wageclaude-sonnet-53/5AI-assisted measurement software reduces technician/physician time on quantification, but licensing costs for validated medical-grade software plus required physician oversight keep costs roughly comparable to human-only workflows in many practices.
Technical feasibility todayclaude-haiku-4-5-202510014/5Multiple FDA-cleared and CE-marked AI products (e.g., by vendors like EchoGo, HealthImageAI, Ultromics) now perform automated cardiac measurement and abnormality detection in production hospital settings with demonstrated reliability. Deployment is growing but not yet universal across all echocardiography labs.
Technical feasibility todayclaude-sonnet-53/5FDA-cleared AI echocardiography tools (e.g., automated ejection fraction and chamber measurement software) are deployed in some cardiology practices, but adoption is uneven and physicians still verify measurements manually in most settings.

Conduct electrocardiogram (EKG), phonocardiogram, echocardiogram, or other cardiovascular tests to record patients' cardiac activity, using specialized electronic test equipment, recording devices, or laboratory instruments.

30

CI 28–32 · exposure 30 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Healthcare organizations have begun adopting AI for EKG and echo interpretation in pilot and production settings, particularly in screening and high-volume environments, but adoption remains uneven and slower than in other sectors due to regulatory and liability concerns. The core acquisition process remains largely manual.
Sector adoption velocityclaude-sonnet-53/5Healthcare/cardiology is adopting AI for image analysis and decision support at a moderate pace, but adoption of full test-conducting automation is slow due to regulatory and clinical workflow constraints.
Augmentation potentialclaude-haiku-4-5-202510014/5AI significantly augments cardiologists and technicians by rapidly flagging abnormalities, quantifying measurements (e.g., ejection fraction), and reducing interpretation time, allowing human experts to focus on complex or borderline cases and clinical decision-making rather than routine signal analysis.
Augmentation potentialclaude-sonnet-54/5AI tools increasingly assist in real-time waveform analysis, arrhythmia detection, and echo image quality feedback, meaningfully augmenting the clinician's efficiency and accuracy during test conduct and interpretation.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in *interpreting* EKG and echocardiogram data with high accuracy, the task also requires hands-on equipment operation, patient positioning, probe placement, and real-time clinical decision-making during the test. Current AI cannot physically perform the test or autonomously decide when additional views or measurements are needed, limiting automation to interpretation alone.
Task automatabilityclaude-sonnet-52/5Test acquisition requires physical patient contact, probe manipulation, and equipment operation that current AI cannot perform end-to-end; AI can assist interpretation but not the hands-on conduct of the test.'
Adoption barriersclaude-haiku-4-5-202510014/5Medical licensing requirements, liability asymmetry (false negatives carry high legal/clinical cost), regulatory oversight (FDA clearance for diagnostic devices), and the clinical requirement for physician sign-off on test results create substantial barriers. Patient contact and hands-on test execution also legally require qualified human operators.
Adoption barriersclaude-sonnet-54/5Physical performance of diagnostic cardiac tests typically requires licensed technicians or physicians, with liability and physical patient-contact requirements limiting substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI interpretation reduces analysis time and can flag abnormalities, but the cost of deploying, maintaining, and clinically validating these systems (plus mandatory physician oversight) remains high relative to a technician's wage. The full workflow—acquisition plus interpretation—still requires human labor at each step.
Cost vs. human wageclaude-sonnet-52/5Specialized equipment and skilled human operators are still required; AI reduces some interpretation time but doesn't eliminate the cost of test administration, so overall cost savings are modest.
Technical feasibility todayclaude-haiku-4-5-202510013/5Deployed products reliably interpret EKG tracings and echocardiograms (e.g., FDA-cleared AI for EKG screening, ejection fraction estimation), but these systems are decision-support tools, not end-to-end test conductors. Human technicians or sonographers still perform the actual data acquisition, and error rates in interpretation remain material for complex cases.
Technical feasibility todayclaude-sonnet-52/5AI-assisted ECG/echo interpretation tools exist in some hospitals, but the actual conducting of tests (positioning, probe handling, image acquisition) is still performed by sonographers/technicians, not AI products.

Advise patients and community members concerning diet, activity, hygiene, or disease prevention.

29

CI 29–29 · exposure 25 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While health systems pilot AI educational tools and decision support, genuine replacement of cardiology advisory consultations remains minimal in production settings. Clinicians integrate AI summaries into workflows rather than displacing themselves, reflecting slow organizational and regulatory adoption of AI autonomy in clinical counseling.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall lags in production AI deployment for patient-facing clinical counseling due to regulatory, liability, and EHR integration hurdles, despite some pilots in patient education tools.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can effectively assist cardiologists by preparing evidence-based talking points, retrieving guideline recommendations, generating patient-friendly educational materials, and flagging risk factors—substantially raising the efficiency and consistency of advisory sessions while the cardiologist retains clinical judgment and rapport.
Augmentation potentialclaude-sonnet-54/5AI can efficiently generate personalized handouts, translate materials, and draft talking points on diet, exercise, and prevention, meaningfully speeding up the physician's counseling prep while the doctor still delivers and tailors the advice.
Task automatabilityclaude-haiku-4-5-202510012/5AI can generate generic health advice and educational content about diet and activity, but cannot assess individual patient risk factors, comorbidities, or compliance barriers that are essential for meaningful medical counseling. The personalization required for effective patient advice remains beyond current AI capability without substantial human judgment.
Task automatabilityclaude-sonnet-52/5General lifestyle and prevention advice can be drafted by AI chatbots, but personalized cardiology counseling requires integrating patient history, risk stratification, and nuanced judgment that current systems cannot fully replicate end-to-end at equal quality.
Adoption barriersclaude-haiku-4-5-202510014/5Medical licensing and liability requirements strongly protect this task: a cardiologist must personally counsel patients on disease management, and AI-generated advice cannot substitute for licensed clinical judgment without legal and malpractice risk exposure. Standard of care and regulatory oversight demand human accountability.
Adoption barriersclaude-sonnet-54/5Clinical advice tied to a patient's diagnosis and treatment plan generally requires a licensed physician for liability and standard-of-care reasons, even though general health education is less restricted.
Cost vs. human wageclaude-haiku-4-5-202510013/5AI-driven educational content and reminders are inexpensive to deploy at scale, making marginal cost per patient contact very low; however, the high-touch advisory component still requires cardiologist time, keeping overall cost per quality advisory interaction comparable to current practice.
Cost vs. human wageclaude-sonnet-53/5AI-generated educational content is very cheap to produce, but the physician's time is largely spent on personalized risk assessment and rapport-building, so overall cost savings are moderate rather than order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510012/5Chatbots and decision-support tools can provide standard educational messaging, but no deployed product reliably performs the full advisory task—which requires understanding the patient's social context, motivations, and medical complexity. Current AI systems lack the clinical judgment needed for production-grade medical advice-giving in cardiology practice.
Technical feasibility todayclaude-sonnet-52/5Consumer health chatbots and symptom checkers exist and give generic diet/lifestyle advice, but no deployed product reliably substitutes for individualized physician counseling in cardiology practice.

Conduct exercise electrocardiogram tests to monitor cardiovascular activity under stress.

28

CI 25–30 · exposure 30 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Hospitals and clinics have adopted AI-assisted ECG interpretation for analysis, but the procedure itself remains a standard clinical workflow with minimal shift toward automation; stress testing still requires direct physician presence and control.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially procedural/physical exam settings, has been slower to adopt full automation compared to administrative or diagnostic-support AI, with pilots more common than production deployment for test administration.
Augmentation potentialclaude-haiku-4-5-202510014/5AI substantially augments cardiologist productivity by providing immediate, accurate interpretation of ECG data during or after the test, highlighting abnormalities and supporting real-time decision-making about test continuation and clinical next steps.
Augmentation potentialclaude-sonnet-54/5AI-assisted ECG analysis tools substantially help cardiologists interpret stress test results faster and flag anomalies, meaningfully boosting diagnostic throughput and accuracy while the physician retains oversight.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can analyze ECG waveforms and detect some arrhythmias with high accuracy, conducting the full exercise test requires real-time patient monitoring, manual electrode placement, test protocol adjustment based on patient response, and clinical judgment to halt testing if complications arise—human involvement is essential for safety and quality.
Task automatabilityclaude-sonnet-52/5AI can assist with signal interpretation and monitoring, but the physical conduct of stress tests requires supervising the patient's physiological response, adjusting protocols, and managing emergencies, which cannot be fully automated today.
Adoption barriersclaude-haiku-4-5-202510014/5Regulatory and liability barriers are substantial: a licensed cardiologist must be present to ensure patient safety during stress testing, interpret findings in real-time, and take responsibility for any adverse events; automation cannot legally substitute for this physician oversight.
Adoption barriersclaude-sonnet-54/5Conducting stress tests involves direct patient physical monitoring and emergency response capability, requiring licensed medical personnel present, with significant liability exposure for cardiac events during testing.
Cost vs. human wageclaude-haiku-4-5-202510012/5ECG analysis AI is inexpensive, but the cardiologist's time remains the dominant cost driver because human presence is required for patient safety monitoring, test administration, and clinical decision-making throughout the procedure.
Cost vs. human wageclaude-sonnet-52/5AI can lower interpretation costs modestly, but the need for on-site physician/technician supervision during physical stress testing keeps overall costs close to current human-staffed models.
Technical feasibility todayclaude-haiku-4-5-202510013/5AI systems can reliably interpret ECG recordings post-hoc and flag abnormalities, and some deployed cardiology platforms include ECG analysis modules; however, no current system autonomously conducts the physical test procedure, patient communication, and on-the-fly protocol modifications needed in a clinical setting.
Technical feasibility todayclaude-sonnet-52/5There are FDA-cleared ECG interpretation algorithms in use, but no product performs the full stress-test procedure (patient supervision, protocol adjustment, safety monitoring) autonomously in production.

Monitor patients' conditions and progress, and reevaluate treatments, as necessary.

28

CI 23–32 · exposure 30 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Healthcare and cardiology practices have piloted AI monitoring tools and dashboard integration, but actual production displacement of cardiologist monitoring remains limited; adoption is cautious due to liability concerns and cultural attachment to physician-directed patient oversight.
Sector adoption velocityclaude-sonnet-53/5Healthcare has growing adoption of AI-assisted monitoring (wearables, remote telemetry, EHR-integrated alerts) but full-scale autonomous reevaluation remains at pilot stage in most cardiology practices.
Augmentation potentialclaude-haiku-4-5-202510014/5AI substantially augments cardiologist productivity by automating data aggregation, flagging abnormal trends in monitoring data, and summarizing imaging or lab changes, enabling faster clinical review and more informed treatment decisions while the cardiologist retains central decision authority.
Augmentation potentialclaude-sonnet-54/5AI-driven monitoring dashboards, predictive alerts, and risk stratification tools meaningfully help cardiologists track patient status and flag when reevaluation is needed, improving efficiency while the physician retains decision authority.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with monitoring via automated analysis of ECGs, imaging, and lab values, the task requires complex clinical judgment to reevaluate treatments based on multifactorial patient context, comorbidities, and nuanced symptom changes that current systems cannot reliably handle end-to-end without substantial human oversight.
Task automatabilityclaude-sonnet-52/5Continuous monitoring and treatment reevaluation require integrating longitudinal clinical judgment, physical exam findings, and patient context that current AI cannot fully synthesize or act upon autonomously.
Adoption barriersclaude-haiku-4-5-202510014/5Strong legal and licensing barriers exist: cardiologists must legally evaluate and authorize treatment changes; malpractice liability and standard-of-care expectations require physician accountability for clinical decisions; and regulatory frameworks (FDA oversight of diagnostic AI) constrain autonomous automation.
Adoption barriersclaude-sonnet-55/5Treatment reevaluation and clinical decision-making for patient care legally require a licensed physician, with strong liability exposure and regulatory oversight (medical licensing, malpractice law) preventing full substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI monitoring systems require infrastructure, integration, and cardiologist oversight costs that offset savings; since human judgment is still required for treatment decisions, the all-in cost remains comparable to or exceeds direct human monitoring labor.
Cost vs. human wageclaude-sonnet-52/5AI monitoring tools add cost for data integration, alert triage, and false-positive management, and still require a cardiologist's paid time to interpret and act, so savings versus the human-in-loop workflow are limited.
Technical feasibility todayclaude-haiku-4-5-202510013/5Deployed products (EHR analytics, AI-assisted ECG interpretation, cardiac imaging tools) exist and perform reliably on narrow components, but no mature system reliably monitors overall patient progress and autonomously reevaluates treatments at production scale without cardiologist review and sign-off.
Technical feasibility todayclaude-sonnet-52/5Deployed products exist for isolated pieces (arrhythmia detection, risk scoring, remote monitoring alerts) but no product independently monitors and reevaluates treatment plans in production without physician oversight.

Diagnose medical conditions of patients, using records, reports, test results, or examination information.

26

CI 20–32 · exposure 30 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Cardiology is a digitized specialty with moderate AI adoption in imaging and monitoring tools, but deployment remains primarily assistive rather than autonomous. Hospitals and practices are pilots-heavy but production replacement of cardiologist diagnosis is rare and cautious.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially clinical diagnosis, is a historically slow-adopting sector for autonomous AI decision-making due to regulation, liability, and integration complexity, though narrow AI tools (imaging, ECG) are seeing steady uptake.
Augmentation potentialclaude-haiku-4-5-202510014/5AI dramatically assists cardiologists by rapidly analyzing imaging, ECG, and lab data, flagging abnormalities, and prioritizing cases—measurably raising productivity and reducing diagnostic time while the cardiologist retains clinical judgment and final decision authority.
Augmentation potentialclaude-sonnet-54/5AI substantially augments diagnosis by flagging abnormalities in ECGs, imaging, and lab data, surfacing relevant literature, and supporting differential diagnosis generation, while the cardiologist retains final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in analyzing cardiac imaging and ECG data with high accuracy, end-to-end diagnosis requires integrating multiple data streams, patient history context, and clinical judgment that current systems struggle to reliably do without significant human oversight. AI tools can flag abnormalities but cannot yet replace the cardiologist's synthesis and decision-making at scale with equal quality and ≥50% time savings.
Task automatabilityclaude-sonnet-52/5Diagnosis requires integrating physical exam findings, patient history, and clinical judgment under uncertainty; AI can support parts (e.g., ECG/imaging interpretation) but cannot yet independently perform full diagnostic workups end-to-end reliably enough for 50%+ time savings at equal quality.
Adoption barriersclaude-haiku-4-5-202510014/5Diagnosis by cardiologists is a licensed, regulated medical function with high liability for errors; malpractice and regulatory frameworks (FDA oversight of diagnostic AI, state medical boards) require physician sign-off and accountability. Strong legal and professional barriers exist to full substitution of the diagnostic act itself.
Adoption barriersclaude-sonnet-55/5Medical diagnosis is a licensed clinical act with strict legal, regulatory (FDA/medical board), and malpractice liability requirements mandating a physician's sign-off, creating a hard barrier to full substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference for cardiac imaging and ECG analysis is cheap, but integration, validation, and the cardiologist oversight required for safe deployment make the all-in cost per diagnosis high relative to a cardiologist's per-task cost. Full automation would still require significant human involvement due to liability and complexity.
Cost vs. human wageclaude-sonnet-52/5AI tools for specific diagnostic components are cheap per-use, but achieving physician-equivalent full diagnostic output requires substantial human oversight, integration, and liability coverage, keeping the effective cost comparable to or higher than incremental physician time.
Technical feasibility todayclaude-haiku-4-5-202510013/5Deployed products (cardiac imaging AI, ECG analysis algorithms) perform well on narrow tasks like detecting specific arrhythmias or valve disease, but comprehensive diagnosis across the full scope of cardiology remains unreliable in production. These tools exist and are used clinically, but typically require cardiologist review and show material error rates on complex cases.
Technical feasibility todayclaude-sonnet-52/5Deployed diagnostic-support tools exist for specific modalities (ECG interpretation, echo analysis, risk scoring) but no product performs comprehensive cardiologic diagnosis autonomously in production; physicians remain the diagnostic decision-maker.

Order medical tests, such as echocardiograms, electrocardiograms, and angiograms.

24

CI 20–28 · exposure 25 · augmentation 63 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While EHR vendors offer clinical decision support modules, adoption of *autonomous* test-ordering AI in cardiology remains minimal and pilot-stage. Most practices still rely on cardiologist manual ordering with light decision-support augmentation.
Sector adoption velocityclaude-sonnet-53/5Healthcare has moderate AI adoption via EHR-integrated clinical decision support and order sets, but full agentic ordering systems are rare and adoption is cautious due to regulation and liability.
Augmentation potentialclaude-haiku-4-5-202510013/5Clinical decision support tools can meaningfully assist by suggesting appropriate tests based on patient data and guidelines, reducing cognitive load and improving consistency. Cardiologists using these tools work faster and with fewer omissions, though they retain final ordering authority.
Augmentation potentialclaude-sonnet-54/5AI-driven clinical decision support and predictive analytics can meaningfully assist cardiologists in identifying which tests are indicated, improving efficiency and diagnostic accuracy while the physician retains final authority.
Task automatabilityclaude-haiku-4-5-202510012/5Ordering tests requires interpreting patient history, symptoms, and prior results to determine clinical necessity—judgment-heavy work that current AI cannot reliably do end-to-end. AI can assist in suggesting test panels based on coded diagnoses, but cannot yet achieve the ≥50% time saving at equal quality threshold without significant human oversight and decision-making.
Task automatabilityclaude-sonnet-52/5Ordering tests requires clinical judgment about differential diagnosis and patient context; AI can suggest appropriate tests but the decision and order authorization remain human-driven with limited full end-to-end automation today.
Adoption barriersclaude-haiku-4-5-202510014/5Ordering tests must be signed by a licensed cardiologist and is medically and legally binding; liability, regulatory oversight (CMS, FDA), and clinical responsibility prevent autonomous substitution. The human-contact requirement is both legal and tied to malpractice exposure.
Adoption barriersclaude-sonnet-55/5Ordering medical tests is a licensed medical act requiring physician authorization under medical practice law and liability frameworks; a non-licensed system cannot legally order tests independently.
Cost vs. human wageclaude-haiku-4-5-202510012/5Clinical decision support and EHR integration costs, plus mandatory oversight by licensed cardiologists, mean automation does not achieve cost advantage. The human remains the rate-limiting step, and the all-in cost of maintaining AI systems plus oversight exceeds the labor-time savings.
Cost vs. human wageclaude-sonnet-52/5AI-assisted decision support adds marginal cost on top of existing EHR systems, but since a licensed physician must still review and authorize, cost savings from AI alone are modest relative to physician time already spent on other tasks.
Technical feasibility todayclaude-haiku-4-5-202510012/5While rule-based clinical decision support systems exist, no deployed product reliably performs independent test ordering without human cardiologist review and final authorization. Current systems operate as narrow, vetted suggestion engines in controlled settings, not autonomous ordering agents.
Technical feasibility todayclaude-sonnet-52/5Clinical decision support tools exist that suggest test ordering based on symptoms/guidelines, but no deployed product autonomously orders cardiac diagnostic tests in production without physician initiation.

Observe ultrasound display screen, and listen to signals to record vascular information, such as blood pressure, limb volume changes, oxygen saturation, and cerebral circulation.

23

CI 21–25 · exposure 25 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Despite advances in AI-assisted ultrasound analysis, adoption in cardiology remains limited to pilot programs and research settings. Most hospital systems still rely on human cardiologists for real-time vascular signal interpretation, with AI integration slow due to regulatory caution and liability concerns.
Sector adoption velocityclaude-sonnet-52/5Healthcare imaging adoption of AI is growing but remains slow and cautious due to regulatory, liability, and workflow integration hurdles compared to faster-moving digital sectors.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can augment cardiologists significantly by highlighting regions of interest in ultrasound, automating routine measurements, flagging abnormal waveforms, and summarizing data—allowing the human to focus on complex interpretation and clinical decision-making while improving speed and consistency.
Augmentation potentialclaude-sonnet-54/5AI-assisted Doppler and image analysis tools meaningfully speed up interpretation and flag abnormalities, enhancing cardiologist throughput and accuracy while the human remains in control of the exam.
Task automatabilityclaude-haiku-4-5-202510012/5Current AI can detect some vascular abnormalities in ultrasound images and waveform analysis, but integrating real-time listening to Doppler signals, dynamic observation, and the clinical judgment required to record actionable vascular information remains beyond autonomous performance today. The task requires continuous adaptive monitoring and contextual interpretation that AI cannot yet replicate reliably end-to-end.
Task automatabilityclaude-sonnet-52/5AI can assist in signal capture and flagging abnormal readings, but real-time observation, probe manipulation, and clinical correlation during the exam still require a human operator; full end-to-end automation is not yet standard practice.
Adoption barriersclaude-haiku-4-5-202510014/5Clinical decision-making based on ultrasound observation carries high liability; diagnostic errors carry direct patient harm. Regulatory bodies (FDA, medical boards) require licensed cardiologists to interpret and take responsibility for vascular assessments, creating strong legal and professional barriers to full automation.
Adoption barriersclaude-sonnet-54/5Medical licensing, liability for diagnostic errors, and regulatory requirements (FDA clearance, physician sign-off) create strong barriers to full automation of this clinical observation task.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of integrating real-time AI monitoring systems (hardware, software, validation, oversight) for this task exceeds the wage of a trained cardiologist performing direct observation, especially when accounting for liability and the need for human verification.
Cost vs. human wageclaude-sonnet-52/5Hardware, sensors, and clinical oversight requirements keep AI-assisted vascular studies costly; savings are incremental (faster reads) rather than order-of-magnitude cost reduction versus a sonographer/cardiologist.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI tools exist for automated ultrasound image analysis and some waveform interpretation, no deployed product reliably performs the full task of real-time vascular signal observation, listening, and comprehensive recording at clinical standards. Existing systems have high error rates on complex cases and narrow scope, falling short of production reliability.
Technical feasibility todayclaude-sonnet-52/5Deployed AI tools exist for image analysis and Doppler waveform interpretation, but reliable autonomous acquisition and interpretation of live vascular studies without a technician/physician is not in production at scale.

Design and explain treatment plans, based on patient information such as medical history, reports, and examination results.

23

CI 20–25 · exposure 30 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Cardiology adopts narrow AI tools (imaging AI, risk scores) steadily but conservatively; end-to-end treatment planning automation has not achieved production adoption because clinical teams and regulators remain cautious about AI designing personalized medical plans without physician oversight.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially high-stakes specialty care like cardiology, adopts AI slowly due to regulatory approval requirements, liability concerns, and integration with EHR systems.
Augmentation potentialclaude-haiku-4-5-202510014/5AI significantly augments cardiologists by rapidly synthesizing patient data, flagging relevant guidelines, suggesting differential diagnoses, and surfacing risk factors, allowing the physician to focus on judgment and patient communication. This is a clear productivity enhancer while the cardiologist retains decision authority.
Augmentation potentialclaude-sonnet-54/5AI can synthesize patient history, flag risk factors, suggest guideline-concordant options, and help draft patient-facing explanations, meaningfully speeding up the cardiologist's planning and communication process.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in analyzing medical data and suggesting evidence-based options, the task of designing a personalized treatment plan requires integration of complex patient context, value judgments, and accountability that current AI cannot reliably perform end-to-end. A cardiologist must weigh trade-offs, consider patient preferences, and assume legal responsibility in ways AI systems today cannot match.
Task automatabilityclaude-sonnet-52/5Designing treatment plans requires integrating nuanced clinical judgment, patient values, and complex risk-benefit tradeoffs that current AI cannot reliably perform end-to-end; explaining plans to patients requires trust and accountability a cardiologist must own.
Adoption barriersclaude-haiku-4-5-202510015/5Cardiology treatment planning is legally and ethically anchored to the licensed physician: liability for adverse outcomes, malpractice exposure, informed consent, and regulatory frameworks (FDA, state medical boards) all require human physician accountability and sign-off. No AI system can legally own the treatment decision.
Adoption barriersclaude-sonnet-55/5Medical licensing laws require a physician to diagnose and authorize treatment plans, and malpractice liability makes autonomous AI treatment planning legally impermissible.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI tools (risk calculators, imaging analysis) reduce some analysis overhead but require significant integration, clinician oversight, and validation; the all-in cost remains high relative to the portion of a cardiologist's time saved, and the loaded wage for a cardiologist's decision-making remains comparatively expensive to displace.
Cost vs. human wageclaude-sonnet-52/5AI decision-support software has licensing and integration costs plus mandatory physician oversight, so total cost per patient encounter is not dramatically cheaper than physician time once liability review is included.
Technical feasibility todayclaude-haiku-4-5-202510013/5Clinical decision support products (e.g., risk stratification tools, diagnostic imaging AI) exist and deploy in cardiology workflows, but they focus on narrow segments (e.g., imaging interpretation, risk calculators) rather than holistic treatment planning. No deployed system reliably designs and explains full personalized treatment plans autonomously.
Technical feasibility todayclaude-sonnet-52/5Clinical decision-support tools exist that suggest guideline-based options, but no deployed product autonomously designs and explains individualized cardiology treatment plans in production without physician authorship.

Answer questions that patients have about their health and well-being.

21

CI 14–29 · exposure 25 · augmentation 63 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Cardiologists and healthcare organizations are highly regulated and cautious; actual adoption of AI for direct patient communication remains minimal in production settings, with most uses limited to triage or informational support under physician oversight.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall adopts AI assistive tools cautiously due to regulation, liability, and patient trust issues, with production use concentrated in administrative and documentation tasks rather than direct patient counseling.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can help cardiologists draft responses, summarize patient questions, or surface relevant clinical guidelines, improving efficiency in communication. However, the physician must review and validate all responses, limiting the multiplicative effect on productivity.
Augmentation potentialclaude-sonnet-54/5AI can help cardiologists draft answers, summarize research, prepare patient education materials, and pre-triage common questions, meaningfully speeding up communication while the physician remains responsible for final answers.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can provide general health information and answer routine questions, cardiologists must answer questions grounded in a specific patient's medical history, test results, and clinical context. This requires synthesis of individual patient data, clinical judgment about risk, and legal/ethical accountability that current AI systems cannot reliably provide end-to-end without substantial human oversight.
Task automatabilityclaude-sonnet-52/5Answering patient health questions requires clinical judgment, contextual knowledge of the patient's history, and accountability for medical advice, which current AI cannot fully replicate end-to-end despite being able to draft or triage some responses.
Adoption barriersclaude-haiku-4-5-202510015/5Medical practice laws require that patient medical advice be provided by a licensed physician; liability for misdiagnosis or harmful guidance, combined with professional licensing requirements and the duty of care standard, create hard legal barriers to full automation or unsupervised AI deployment.
Adoption barriersclaude-sonnet-54/5Medical advice is subject to licensing, liability, and malpractice exposure, and many patients expect a licensed physician to address their specific health concerns, creating strong regulatory and trust barriers.
Cost vs. human wageclaude-haiku-4-5-202510011/5A cardiologist's time is highly valued ($200–300k+ annual salary); AI infrastructure plus required human review and liability coverage does not yet achieve cost advantage for this safety-critical, patient-specific task.
Cost vs. human wageclaude-sonnet-53/5AI can cheaply generate draft answers to common questions, but the need for physician oversight and liability review narrows the cost advantage compared to a fully autonomous system.
Technical feasibility todayclaude-haiku-4-5-202510012/5Chatbots and LLMs can answer generic health questions, but no deployed product reliably handles patient-specific medical queries with the accuracy and accountability required in a clinical setting. Products exist but cannot substitute for a cardiologist's contextual judgment without material legal and safety risks.
Technical feasibility todayclaude-sonnet-52/5Patient-facing chatbots and portals exist for basic FAQs and triage, but no deployed product reliably handles nuanced clinical Q&A from cardiologists to patients without physician review.

Conduct tests of the pulmonary system, using a spirometer or other respiratory testing equipment.

19

CI 13–25 · exposure 17 · augmentation 50 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare settings are adopting AI for interpretation of existing test data, but the physical test administration itself has seen minimal AI displacement because of licensure requirements and the need for real-time patient interaction.
Sector adoption velocityclaude-sonnet-52/5Healthcare diagnostic testing environments adopt automation slowly due to regulatory, safety, and clinical validation requirements, despite general AI advances in other sectors.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by automatically flagging abnormal patterns in real-time spirometry results and recommending repeat tests, helping clinicians interpret findings more efficiently while they remain responsible for test administration.
Augmentation potentialclaude-sonnet-53/5AI can assist in analyzing spirometry curves, flagging abnormalities, and supporting differential diagnosis, providing moderate productivity gains for the clinician interpreting results.
Task automatabilityclaude-haiku-4-5-202510011/5Spirometry and pulmonary function testing require hands-on manipulation of specialized respiratory equipment and direct physical interaction with patients to obtain valid measurements. Current AI systems cannot physically operate this equipment or guide patient breathing techniques in real time.
Task automatabilityclaude-sonnet-52/5Performing the physical spirometry test requires patient interaction, equipment setup, and coaching effort that current AI cannot execute; AI can assist with interpretation but not the hands-on test administration.
Adoption barriersclaude-haiku-4-5-202510014/5Clinical testing requires a licensed healthcare professional to administer and validate measurements; liability and regulatory oversight (FDA, CLIA) mandate human certification and responsibility for test quality and patient safety.
Adoption barriersclaude-sonnet-54/5Diagnostic pulmonary function testing is clinically regulated, requires trained personnel for proper technique and calibration, and results feed into physician-signed diagnoses, creating strong procedural and liability barriers.
Cost vs. human wageclaude-haiku-4-5-202510011/5The labor cost of a trained technician or cardiologist administering spirometry is modest relative to the capital equipment and liability exposure, and AI cannot replace the human performing the test itself.
Cost vs. human wageclaude-sonnet-52/5The equipment itself is not AI-driven labor replacement; a technician or nurse must still be present, so cost savings versus human labor are minimal despite automated readouts.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI can assist in interpreting spirometry results from existing data, but no deployed product performs the actual hands-on testing and equipment operation. Production systems exist only for result analysis, not test administration.
Technical feasibility todayclaude-sonnet-52/5Automated spirometers exist and can auto-calculate metrics, but a human (technician or clinician) still administers the test, ensures proper technique, and validates results in production settings.

Explain procedures and discuss test results or prescribed treatments with patients.

19

CI 14–24 · exposure 25 · augmentation 63 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Despite healthcare's digital transformation, autonomous AI-led patient communication remains deeply restricted by liability, regulation, and professional norms; adoption of AI for this specific task in production is minimal and unlikely to accelerate without fundamental changes to medical licensing and liability law.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall adopts AI slowly for direct patient communication due to regulatory, liability, and trust concerns, though administrative and documentation uses are growing faster.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist cardiologists by drafting clear explanations, summarizing complex test results, or preparing talking points, raising the efficiency and consistency of the human-delivered communication, but the physician remains the primary communicator and decision-maker.
Augmentation potentialclaude-sonnet-54/5AI can help cardiologists draft patient-friendly explanations, summarize test results, and prepare talking points, meaningfully improving efficiency and communication quality while the physician remains the primary communicator.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can generate explanations of procedures and summarize test results from medical records, the task requires individualized patient communication, empathy, and real-time responsiveness to patient concerns and questions that current AI systems cannot reliably handle end-to-end. Significant human oversight and revision would be needed, falling well short of the 50% time-saving threshold for unmodified autonomous performance.
Task automatabilityclaude-sonnet-52/5While AI chatbots can generate explanations of medical procedures and test results, the interactive, empathetic, patient-specific discussion requiring clinical judgment and trust-building cannot be fully automated end-to-end today.
Adoption barriersclaude-haiku-4-5-202510015/5This task carries hard legal and regulatory barriers: a licensed physician must directly communicate with and obtain informed consent from the patient; malpractice liability rests with the physician, not an AI system; and patient contact is explicitly required by medical practice standards and law.
Adoption barriersclaude-sonnet-55/5Discussing diagnoses, test results, and treatment plans is a core licensed medical act with strong informed-consent and liability requirements, legally requiring physician involvement.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of AI systems (inference, integration, compliance infrastructure, and mandatory physician review/revision time) exceeds the time savings, since a cardiologist must remain in the loop and ultimately deliver the explanation themselves to meet legal and quality requirements.
Cost vs. human wageclaude-sonnet-53/5AI-generated explanatory content is cheap to produce, but the physician's time for discussion and liability-bearing communication still dominates cost, keeping overall ratio roughly comparable when quality and trust are factored in.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs this task independently in production. AI systems can draft explanations or summaries, but patient-facing medical communication requires licensed physician judgment, informed-consent documentation, and handling of patient emotions—all areas where current AI fails to meet clinical and legal standards for autonomous use.
Technical feasibility todayclaude-sonnet-52/5Patient-facing AI tools (e.g., portals with AI-generated summaries) exist but are largely supplementary; no deployed product reliably conducts the full physician-patient treatment discussion in production.

Conduct research to develop or test medications, treatments, or procedures that prevent or control disease or injury.

14

CI 3–25 · exposure 13 · augmentation 63 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While research institutions use AI tools for analysis, the pace of AI-driven autonomous medication/procedure development and testing remains slow; most adoption is assistive rather than substitutive, and actual deployment of AI-generated treatments still requires extensive human validation.
Sector adoption velocityclaude-sonnet-52/5Medical research is a highly regulated, slow-moving sector where AI adoption for actual trial conduct remains at pilot/tool-assist stage rather than production-scale autonomous research.
Augmentation potentialclaude-haiku-4-5-202510013/5AI meaningfully assists cardiologists in literature synthesis, data analysis, and pattern recognition across research datasets, but the core creative and validation work remains human-driven; augmentation is useful but not transformative for the full research pipeline.
Augmentation potentialclaude-sonnet-54/5AI substantially assists literature review, hypothesis generation, data analysis, and drug candidate screening, meaningfully boosting researcher productivity while humans retain control.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in literature review, data analysis, and hypothesis generation for drug/treatment research, the core experimental design, hands-on testing, clinical validation, and judgment-driven decision-making in developing or testing new procedures require human expertise and cannot be fully automated end-to-end today.
Task automatabilityclaude-sonnet-51/5Original clinical research including hypothesis generation, trial design, patient enrollment, ethical oversight, and physical experimentation cannot be executed end-to-end by current AI systems.assistive.
Adoption barriersclaude-haiku-4-5-202510014/5Regulatory requirements (FDA approval pathways, IRB oversight, clinical trial regulations) legally mandate human researcher responsibility and clinical decision-making; liability for experimental outcomes and patient safety creates strong institutional and legal barriers to full automation.
Adoption barriersclaude-sonnet-55/5Clinical research requires IRB approval, regulatory oversight (FDA/EMA), licensed physician involvement, and legal accountability for patient safety, making full automation legally and ethically barred.
Cost vs. human wageclaude-haiku-4-5-202510012/5Conducting original research, especially clinical trials and procedure development, involves expensive infrastructure, regulatory compliance, and specialized human oversight that far exceeds current AI inference costs; AI augments but does not replace the expensive human-led process.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the full research process, so there is no meaningful cost-equivalent comparison; human-led research with AI tools still requires the same expensive infrastructure and personnel.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI tools exist for data analysis and literature mining in research contexts, but no deployed product autonomously conducts the full research cycle—from experimental design through procedure/medication testing to clinical validation—at production scale in cardiology.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously conducts cardiology research from design through validated conclusions; AI is used only as a tool within human-led research.

Recommend surgeons or surgical procedures.

13

CI 6–20 · exposure 13 · augmentation 50 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Despite general interest in AI in healthcare, actual adoption of autonomous surgical recommendation systems in cardiology practice is minimal; most implementations remain decision-support tools with human physicians retaining full decision authority.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall adopts AI slowly for high-stakes clinical decisions, with adoption concentrated in diagnostic imaging and documentation rather than procedural recommendations.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist cardiologists by summarizing clinical evidence, highlighting imaging findings, and surfacing guideline-aligned options, but the human cardiologist remains the essential decision-maker integrating patient values and clinical judgment.
Augmentation potentialclaude-sonnet-53/5AI can help synthesize evidence, surgical outcomes data, and guidelines to support cardiologists' decision-making, though the final recommendation remains human-driven.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist in summarizing clinical guidelines and generating candidate surgical options based on imaging and test results, but the task fundamentally requires human clinical judgment integrating patient-specific factors, comorbidities, and risk assessment that current systems cannot reliably do end-to-end with 50% time savings at equal quality.
Task automatabilityclaude-sonnet-51/5Recommending surgeons/procedures requires integrating clinical judgment, patient-specific risk factors, and relational trust that current AI cannot end-to-end replicate reliably.
Adoption barriersclaude-haiku-4-5-202510015/5Medical malpractice liability, regulatory oversight (FDA, state medical boards), and the legal requirement for a licensed physician to take clinical responsibility for surgical recommendations create hard legal barriers to autonomous AI recommendation in clinical practice.
Adoption barriersclaude-sonnet-55/5This is a licensed medical judgment task with direct liability implications; only a physician can legally make and be accountable for such recommendations.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI clinical decision support requires significant integration, validation, and cardiologist oversight, making it comparable to or more expensive than a cardiologist simply making the recommendation directly without AI assistance.
Cost vs. human wageclaude-sonnet-52/5While AI-assisted decision support may be cheap per query, the liability and oversight costs of using AI for surgical recommendations keep effective cost comparable to or higher than physician time.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI tools exist for clinical decision support and imaging analysis, no deployed product reliably makes independent surgical recommendations in production cardiology settings; human cardiologists must still validate all suggestions and integrate complex clinical context.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously makes surgeon or procedure recommendations in clinical practice; decision-support tools exist only as adjuncts to physician judgment.

Prescribe heart medication to treat or prevent heart problems.

13

CI 3–23 · exposure 13 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Healthcare has adopted clinical decision-support tools at moderate pace, but autonomous or AI-led prescribing remains rare. Hospitals and practices pilot such tools, but conservative regulation and liability concerns slow deep production adoption in actual prescription workflows.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially direct clinical prescribing, adopts AI cautiously due to regulation, liability, and safety concerns, despite faster adoption in diagnostics or documentation.
Augmentation potentialclaude-haiku-4-5-202510014/5AI assistants (evidence summaries, drug-interaction checkers, guideline-based recommendations) substantially augment a cardiologist's efficiency in reviewing options and building a prescription, even as the physician retains final authority and responsibility.
Augmentation potentialclaude-sonnet-54/5AI decision-support tools can meaningfully assist cardiologists by flagging drug interactions, suggesting dosing based on guidelines, and summarizing patient data, improving prescribing efficiency and safety.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist in identifying relevant medications based on clinical guidelines and patient data, but prescribing involves nuanced judgment about contraindications, drug interactions, comorbidities, and individual patient factors that require licensed physician oversight. Current AI cannot reliably handle the full task end-to-end at 50% time savings while maintaining safety standards.
Task automatabilityclaude-sonnet-51/5Prescribing requires integrated clinical judgment, patient-specific risk assessment, and legal accountability that current AI cannot autonomously perform end-to-end.'
Adoption barriersclaude-haiku-4-5-202510015/5Prescribing is legally restricted to licensed physicians; a cardiologist must write and take legal responsibility for all prescriptions. Regulatory bodies (FDA, DEA, medical boards) mandate human professional judgment and licensure, creating hard barriers to full automation.
Adoption barriersclaude-sonnet-55/5Prescribing is tightly regulated; only licensed physicians (or authorized prescribers) can legally prescribe medications, creating a hard legal barrier to automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-powered clinical decision support is relatively inexpensive per query, but the human cardiologist remains essential and non-substitutable, so total cost savings are minimal. Integration, validation, and liability oversight add significant overhead without replacing the physician cost.
Cost vs. human wageclaude-sonnet-51/5Since AI cannot legally or reliably perform this task alone, there is no valid cost comparison—human physician cost remains mandatory.
Technical feasibility todayclaude-haiku-4-5-202510012/5Clinical decision-support tools exist and assist with medication recommendations, but no deployed product autonomously prescribes cardiac medications without physician review and sign-off. Products perform narrowly (e.g., flagging drug interactions) rather than end-to-end prescription generation.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently prescribes cardiac medications; clinical decision support tools exist but require physician sign-off, remaining research/assistive stage for autonomous prescribing.

Operate diagnostic imaging equipment to produce contrast-enhanced radiographs of heart and cardiovascular system.

7

CI 0–14 · exposure 13 · augmentation 50 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Cardiac imaging equipment operation is performed by credentialed technicians in clinical settings with stringent regulatory oversight; adoption of autonomous operation is minimal because regulatory and safety requirements remain fixed.
Sector adoption velocityclaude-sonnet-51/5Interventional cardiology procedures remain highly manual and hands-on; AI adoption in this specific physical equipment-operation task is essentially nonexistent in production settings.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by suggesting optimal imaging protocols, flagging acquisition quality issues, or recommending parameter adjustments, but the human operator remains essential for safe, compliant equipment use and real-time decision-making.
Augmentation potentialclaude-sonnet-53/5AI can assist with image reconstruction, enhancement, and post-processing analysis of the resulting radiographs, but does not meaningfully help with the physical equipment operation and contrast administration step itself.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can analyze completed cardiac imaging, the task of *operating* diagnostic equipment—positioning patients, selecting parameters, timing contrast injection, and adjusting settings in real-time—remains primarily manual and requires continuous human judgment and physical intervention that current systems cannot replace end-to-end.
Task automatabilityclaude-sonnet-51/5Physically operating catheterization/fluoroscopy equipment and administering contrast during invasive angiography requires hands-on manipulation of medical devices and patient management that current AI cannot perform.'
Adoption barriersclaude-haiku-4-5-202510015/5Regulatory bodies (FDA, ACR) require licensed technicians or physicians to operate diagnostic imaging equipment and maintain quality control; liability for improper imaging technique is substantial and legally assigned to qualified humans.
Adoption barriersclaude-sonnet-55/5Operating invasive cardiac imaging equipment requires licensed physician/technologist credentials, sterile procedure protocols, and direct liability for patient safety, making substitution legally and physically barred.
Cost vs. human wageclaude-haiku-4-5-202510011/5The capital cost of diagnostic imaging equipment and the specialized training required to operate it safely far exceed any current AI alternative; automation of operation itself is not a cost-competitive option.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing the equipment operation itself, so AI cost is not comparable; human labor plus equipment is the only current option.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI products exist for *interpreting* cardiac images post-acquisition, but no deployed system autonomously operates imaging equipment itself; equipment operation remains a technician or physician function in production settings.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously operates imaging equipment or performs invasive contrast procedures; this remains squarely a physician/technologist-performed physical task.

Administer emergency cardiac care for life-threatening heart problems, such as cardiac arrest and heart attack.

4

CI 0–7 · exposure 5 · augmentation 50 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Emergency cardiac care remains deeply human-centric and physically bound. Adoption of AI in this domain is limited to decision support, not autonomous intervention, and healthcare remains a laggard sector for core clinical task automation.
Sector adoption velocityclaude-sonnet-52/5While hospitals use AI for diagnostics and monitoring (e.g., ECG interpretation), adoption of AI in actual emergency intervention delivery remains minimal and cautious due to patient safety stakes.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist cardiologists by rapidly interpreting ECGs, suggesting protocols, and monitoring vital signs in real-time, which may improve decision speed and reduce errors. However, the human clinician must remain in control of all interventions and clinical judgment.
Augmentation potentialclaude-sonnet-53/5AI tools can assist by flagging abnormal rhythms, predicting risk, and supporting triage decisions, improving speed and accuracy of human-led emergency response.
Task automatabilityclaude-haiku-4-5-202510011/5Emergency cardiac care requires immediate physical intervention (defibrillation, intubation, medication administration, chest compressions) and real-time clinical judgment in chaotic, variable conditions. Current AI cannot perform these interventions end-to-end or achieve 50% time savings on the full task.
Task automatabilityclaude-sonnet-51/5Emergency cardiac care requires immediate hands-on physical intervention (defibrillation, CPR, catheterization, medication administration) that current AI cannot physically perform or independently direct.
Adoption barriersclaude-haiku-4-5-202510015/5Legal and regulatory requirements strictly mandate that licensed physicians and credentialed healthcare providers personally administer emergency cardiac care. Liability, patient consent, and scope-of-practice laws create hard barriers to automation.
Adoption barriersclaude-sonnet-55/5Emergency medical intervention requires licensed physicians and clinical teams, with strict legal, liability, and regulatory requirements mandating human authorization and hands-on care.
Cost vs. human wageclaude-haiku-4-5-202510011/5Cardiologists command high wages ($400k+/year loaded), and AI cannot reduce the cost of emergency care delivery below human cost because physical interventions must still be performed by licensed clinicians or trained paramedics.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this task, so cost comparison favors the human entirely; the physical and time-critical nature makes AI infeasible as a replacement at any cost.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can assist with ECG interpretation and triage algorithms in controlled settings, no deployed product reliably handles the full scope of emergency cardiac care delivery. Clinical decision support exists but falls far short of autonomous care administration.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously administers emergency cardiac care; AI exists only in adjunct roles like arrhythmia detection or decision support, not as an actor performing the intervention.

Talk to other physicians about patients to create a treatment plan.

3

CI 0–6 · exposure 0 · augmentation 50 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5This task sits at the core of physician autonomy and professional responsibility; healthcare adoption of AI for autonomous clinical decision-making and inter-provider communication remains minimal and faces strong institutional and legal resistance.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall shows slow, cautious AI adoption for core clinical decision-making, especially for interpersonal consultations between physicians, due to regulatory and liability constraints.
Augmentation potentialclaude-haiku-4-5-202510012/5AI might assist by summarizing patient data or flagging relevant literature before a physician-to-physician discussion, but the consultation itself—where clinical judgment, accountability, and collegial deliberation occur—is inherently human and resists augmentation.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by synthesizing patient data, summarizing history, flagging drug interactions, or suggesting evidence-based options that inform the physicians' discussion, though it doesn't replace the dialogue itself.
Task automatabilityclaude-haiku-4-5-202510011/5This task fundamentally requires real-time, bidirectional dialogue between licensed physicians to negotiate complex clinical decisions and accountability. Current AI systems cannot participate as peers in medical decision-making or assume responsibility for treatment recommendations.
Task automatabilityclaude-sonnet-51/5This is a synchronous, judgment-heavy clinical collaboration requiring real-time reasoning, negotiation, and accountability between licensed physicians; no AI system can conduct this end-to-end today.2
Adoption barriersclaude-haiku-4-5-202510015/5Physicians have legal and ethical obligations to exercise independent clinical judgment; treatment planning must be documented as a deliberate clinical decision by licensed providers. Liability and regulatory requirements prohibit outsourcing this cognitive and fiduciary responsibility to AI systems.
Adoption barriersclaude-sonnet-55/5Treatment planning is a core licensed medical function with direct liability and legal accountability, requiring physician judgment and sign-off that cannot be delegated to AI.
Cost vs. human wageclaude-haiku-4-5-202510012/5Even if AI could draft summaries or suggestions, the physician time required to review, validate, and discuss with colleagues would remain substantial, offering minimal cost savings over direct conversation.
Cost vs. human wageclaude-sonnet-51/5Because AI cannot perform this interpersonal, liability-bearing task, there is no viable AI-only cost comparison; the human interaction remains mandatory.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product can reliably substitute for physician-to-physician consultation or generate treatment plans that physicians would legally accept as a replacement for direct clinical discussion and collegial review.
Technical feasibility todayclaude-sonnet-51/5No deployed product substitutes for physician-to-physician treatment planning discussions; at most AI tools summarize records or suggest options as background support, not replace the interaction.

Diagnose cardiovascular conditions, using cardiac catheterization.

3

CI 0–5 · exposure 5 · augmentation 50 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Adoption of AI for autonomous cardiac catheterization is not occurring; the invasive and high-risk nature of the task, combined with strict licensure requirements, prevents any meaningful displacement or automation at scale in clinical practice.
Sector adoption velocityclaude-sonnet-51/5Interventional cardiology remains a highly manual, physically-performed specialty with minimal AI-driven procedural adoption; imaging-assist tools are used sparingly and adoption is slow for the procedural task itself.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist with real-time image analysis during catheterization (e.g., vessel tracking, lesion detection on fluoroscopy), and post-procedure quantitative analysis, moderately supporting the cardiologist's workflow without replacing the procedure itself.
Augmentation potentialclaude-sonnet-53/5AI can assist with pre-procedure risk assessment, image interpretation (e.g., angiography analysis), and decision support, meaningfully aiding diagnosis even though it cannot perform the catheterization.
Task automatabilityclaude-haiku-4-5-202510011/5Cardiac catheterization is an interventional procedure requiring real-time manual dexterity, spatial navigation of catheters through vasculature, and immediate clinical judgment in response to live physiological signals. Current AI systems cannot perform the invasive procedural component or make the split-second decisions required during the intervention.
Task automatabilityclaude-sonnet-51/5Cardiac catheterization is an invasive physical procedure requiring manual dexterity, real-time judgment, and direct patient contact; no AI system today can perform the procedure itself end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Cardiac catheterization is explicitly regulated and must be performed by a licensed physician; there is no legal pathway for an AI system to conduct the procedure autonomously, and malpractice liability for any adverse outcome remains with the credentialed clinician.
Adoption barriersclaude-sonnet-55/5Performing cardiac catheterization legally requires a licensed physician with specialized training; strict regulatory, liability, and safety requirements make this a hard-barrier task.
Cost vs. human wageclaude-haiku-4-5-202510011/5The procedure requires expensive specialized equipment, a trained cardiologist's direct involvement, and cannot be meaningfully replaced by AI inference costs. The loaded cost of the human operator (including training, malpractice, and facility overhead) currently sets a floor that AI tools cannot undercut for this invasive task.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute for the physical procedure, so cost comparison is moot—human interventional cardiologists remain the only option, making AI effectively far more 'expensive' (infeasible) for the core task.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI can assist with pre-procedure imaging analysis and post-procedure data interpretation, but no deployed system independently performs cardiac catheterization. Research prototypes exist for catheter guidance, but they require human operators and lack the reliability and regulatory clearance for autonomous use.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs catheterization or the full diagnostic act; AI is at most used for image analysis support within a human-led procedure.

Inject contrast media into patients' blood vessels.

3

CI 0–5 · exposure 5 · augmentation 25 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Adoption of autonomous injection systems in cardiology is nascent; the overwhelming majority of contrast injections are performed by human cardiologists, and organizational inertia in healthcare favors established procedural protocols.
Sector adoption velocityclaude-sonnet-51/5Healthcare procedural tasks involving direct physical patient contact show minimal AI displacement; robotics/automation adoption for such invasive steps is essentially nonexistent in mainstream practice.
Augmentation potentialclaude-haiku-4-5-202510012/5AI can assist cardiologists through image analysis and procedure planning, but the core manual task of injection itself offers limited augmentation potential; the human physician's steady hand and real-time adjustment remain irreplaceable.
Augmentation potentialclaude-sonnet-52/5AI can assist with dosing calculations, protocol selection, or monitoring for adverse reactions, but it provides little direct augmentation to the physical act of injection itself.
Task automatabilityclaude-haiku-4-5-202510011/5Injecting contrast media into blood vessels requires physical manipulation of needles, catheters, and syringes in a patient's body—a task entirely dependent on robotic surgical systems that current AI alone cannot control reliably. While AI can assist in imaging and planning, it cannot perform the procedural injection itself without specialized hardware not yet in clinical deployment.
Task automatabilityclaude-sonnet-51/5This is a hands-on invasive medical procedure requiring physical manipulation of catheters/needles and real-time judgment about patient anatomy and reaction; no AI system can physically perform vascular injection today.
Adoption barriersclaude-haiku-4-5-202510015/5Regulatory bodies (FDA, medical boards) require a licensed physician to perform and take responsibility for invasive vascular procedures. Patient safety liability and the need for real-time human judgment and responsiveness create strict legal and professional barriers to full automation.
Adoption barriersclaude-sonnet-55/5Injecting contrast media is an invasive medical act requiring licensure, direct human physical presence, and legal/clinical liability for adverse reactions, making it a hard-barrier task by definition.
Cost vs. human wageclaude-haiku-4-5-202510011/5Implementing surgical robotics for this task would require capital expenditure on hardware, software, training, and ongoing maintenance—vastly exceeding the direct cost of a cardiologist performing the injection manually.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical task, so the cost comparison is moot—AI cannot deliver this output at any cost, making it effectively infinitely more 'expensive' than the human alternative.
Technical feasibility todayclaude-haiku-4-5-202510012/5Robotic-assisted interventional procedures exist in research settings, but no deployed clinical product reliably performs autonomous contrast injection in routine practice. Human cardiologists remain the standard of care for this procedure.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs contrast injection into patients; this remains a purely manual clinical procedure performed by physicians or trained staff.

Perform vascular procedures, such as balloon angioplasty and stents.

3

CI 0–5 · exposure 5 · augmentation 50 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Despite decades of robotics research, adoption of autonomous or semi-autonomous systems in interventional cardiology remains minimal in practice. The high stakes, regulatory scrutiny, and physician liability concerns have kept adoption velocity very low.
Sector adoption velocityclaude-sonnet-51/5Interventional cardiology remains a highly manual, physically-mediated specialty with minimal automation penetration; adoption of AI is limited to imaging/decision support, not procedural execution.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist by analyzing pre-procedure imaging, suggesting lesion characteristics, and providing intraoperative guidance overlays, but the cardiologist remains the hands-on decision-maker and operator throughout the procedure.
Augmentation potentialclaude-sonnet-53/5AI-enhanced imaging, navigation, and robotic-assisted catheter systems can improve precision and reduce radiation exposure, but the physician performs and controls the procedure throughout.
Task automatabilityclaude-haiku-4-5-202510011/5Vascular procedures like balloon angioplasty and stents require real-time manual dexterity, three-dimensional spatial reasoning in a live patient, and immediate adaptive decision-making based on tactile and visual feedback. Current AI systems cannot perform the physical manipulation or the dynamic intraoperative judgment required.
Task automatabilityclaude-sonnet-51/5This is a hands-on invasive surgical procedure requiring physical manipulation of catheters within blood vessels; no current AI system can perform this end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Vascular procedures require licensure as a physician and specialized interventional training; liability for patient harm is severe and asymmetric; regulatory bodies (FDA, medical boards) mandate human physician control over invasive vascular interventions. Legal and safety barriers are essentially absolute.
Adoption barriersclaude-sonnet-55/5Performing invasive vascular procedures legally requires a licensed physician with specialized training, plus strict liability, sterility, and regulatory requirements around medical device use.
Cost vs. human wageclaude-haiku-4-5-202510011/5The combined cost of interventional equipment, real-time monitoring, robotic systems (if pursued), and the liability and oversight infrastructure would far exceed the loaded cost of a trained interventional cardiologist performing the procedure.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing the physical procedure, so cost comparison favors the human interventionalist entirely; robotic assistance systems add cost rather than reduce it.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can assist with pre-procedure imaging analysis and planning, no deployed system performs the actual catheter placement, balloon inflation, or stent deployment autonomously. Research prototypes exist but lack the safety validation and real-time responsiveness needed for clinical use.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously performs angioplasty or stent placement; robotic-assisted systems exist but require full physician control and are not autonomous.

Perform minimally invasive surgical procedures, such as implanting pacemakers and defibrillators.

1

CI 0–3 · exposure 0 · augmentation 50 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Robotic-assisted cardiac surgery adoption is slow and limited to high-resource hospitals; true autonomous surgical AI adoption in cardiac procedures remains negligible because the technology does not yet exist and regulatory pathways are extremely restrictive.
Sector adoption velocityclaude-sonnet-51/5Surgical specialties involving invasive procedures show minimal AI-driven automation in production; adoption is confined to imaging/planning support, not the physical procedure itself.
Augmentation potentialclaude-haiku-4-5-202510013/5Robotic surgical platforms can assist cardiologists by providing magnification, tremor reduction, and ergonomic support during implantation procedures, modestly improving precision and reducing fatigue, though the surgeon remains fully in control.
Augmentation potentialclaude-sonnet-53/5AI and robotic-assisted systems, image guidance, and procedural planning tools can enhance precision and preoperative decision-making, but the core surgical act remains human-performed.
Task automatabilityclaude-haiku-4-5-202510011/5Minimally invasive cardiac surgery requires real-time haptic feedback, spatial reasoning in a living patient's body, and adaptive decision-making in response to unpredictable anatomical variations and complications. Current AI systems cannot perform autonomous surgical manipulation or replace the surgeon's tactile and visual judgment required for safe device implantation.
Task automatabilityclaude-sonnet-51/5This is a hands-on invasive surgical procedure requiring physical dexterity, real-time tactile feedback, and judgment; no current AI system can perform the physical implantation of a device end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Cardiac surgery is strictly regulated; a licensed physician must perform the procedure and bear legal responsibility for patient outcomes. Medical licensing, malpractice liability, and patient safety requirements create insurmountable legal and regulatory barriers to autonomous automation.
Adoption barriersclaude-sonnet-55/5Only licensed physicians (cardiologists/electrophysiologists) may legally perform invasive cardiac procedures, with strict regulatory, credentialing, and liability requirements making substitution essentially impossible.
Cost vs. human wageclaude-haiku-4-5-202510011/5The infrastructure, training, validation, and oversight costs for any surgical automation system far exceed the hourly cost of an experienced cardiologist, and regulatory requirements would add substantial compliance overhead.
Cost vs. human wageclaude-sonnet-51/5There is no AI system performing this task, so no cost comparison favoring AI exists; the physical procedure still requires a full surgical team and equipment costs equal or exceeding human-only costs.
Technical feasibility todayclaude-haiku-4-5-202510011/5While robotic surgical platforms exist and can assist surgeons, no AI system today operates autonomously or end-to-end to perform pacemaker or defibrillator implantation without direct human control. Surgical robots remain surgeon-controlled tools rather than autonomous agents.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously performs pacemaker or defibrillator implantation; robotic assistance exists in adjacent fields but not as an autonomous replacement for this specific procedure.

Supervise or train cardiology technologists or students.

1

CI 0–3 · exposure 0 · augmentation 38 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Academic medical centers and cardiology programs remain traditional in supervision and training practices; AI adoption in clinical education is nascent, with humans remaining the primary supervisors and trainers across deployed settings.
Sector adoption velocityclaude-sonnet-52/5Healthcare education and clinical supervision remain slow to adopt AI-driven substitutes due to regulatory, liability, and accreditation constraints, though digital training aids are increasingly used as supplements.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist with content generation, scheduling, and record-keeping for training programs, but the core task of evaluating and mentoring trainees relies fundamentally on human judgment, making meaningful augmentation limited.
Augmentation potentialclaude-sonnet-53/5AI-based simulation tools, case libraries, and adaptive learning platforms can meaningfully support training content delivery and knowledge assessment, augmenting but not replacing the cardiologist's supervisory role.
Task automatabilityclaude-haiku-4-5-202510011/5Supervision and training of personnel fundamentally requires human judgment, mentorship, and adaptive feedback tailored to individual learners. AI cannot meaningfully replace the interpersonal and evaluative aspects core to this task.
Task automatabilityclaude-sonnet-51/5Supervising and training cardiology technologists or students requires live clinical judgment, mentorship, hands-on demonstration, and adaptive feedback that AI cannot replicate end-to-end today.
Adoption barriersclaude-haiku-4-5-202510015/5Accreditation standards and licensing requirements mandate that qualified physicians supervise and train technologists and trainees; regulatory and professional standards legally require human oversight of clinical education and competency validation.
Adoption barriersclaude-sonnet-55/5Supervision and certification of trainees is tied to licensure, accreditation standards, and legal responsibility for patient safety, requiring a credentialed physician to oversee training and sign off on competency.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of AI-assisted oversight and training (including LLM-based tutoring, simulation, and content generation) does not yet reduce the need for a licensed cardiologist to verify and sign off on training quality, leaving the loaded cardiologist wage as the binding cost.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this supervisory/training function, so cost comparison favors the human role entirely.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product can supervise or train human technologists or students reliably; this task requires real-time interpersonal interaction, assessment of competency, and adaptive corrective feedback that current AI systems cannot provide in practice.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs supervisory or clinical training roles for cardiology staff; this remains firmly a human responsibility with only ancillary tools (e.g., e-learning modules) available.

Related occupations — Healthcare Practitioners & Technical

How to read this

A high substitution score does not mean this job disappears — it means a large share of its current tasks face replacement pressure, so the mix of tasks is likely to change. High augmentation alongside substitution typically means the occupation reorganizes around the protected tasks. Wide confidence intervals mean the rater panel disagreed: treat those scores as open questions, not verdicts.

What would change this score

New model capabilities (automatability, feasibility), falling inference costs (cost ratio), regulation and licensing shifts (barriers), and measured sector adoption (velocity) all re-enter at every index release. Each release is recomputed, versioned and kept queryable — scores are claims with a date on them, not permanent labels.