Medical Dosimetrists

29-2036.00

Median wage $147,470/yr3,410 employed (US)Rank #492 of 923 scored · top 53% by substitution

Generate radiation treatment plans, develop radiation dose calculations, communicate and supervise the treatment plan implementation, and consult with members of radiation oncology team.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution27

Exposure30

Augmentation66

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

19 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

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%31

panel mean rating 2.2/5 → substitution pressure 31/100

Technical feasibility todayw 20%29

panel mean rating 2.2/5 → substitution pressure 29/100

Cost vs. human wagew 15%28

panel mean rating 2.1/5 → substitution pressure 28/100

Adoption barriersw 20%inverted — strong barriers lower the score19

panel mean rating 4.2/5 (barrier strength) → substitution pressure 19/100

Sector adoption velocityw 10%29

panel mean rating 2.2/5 → substitution pressure 29/100

Task breakdown (19 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.

Record patient information, such as radiation doses administered, in patient records.

CI 62–74 · exposure 70 · augmentation 88 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Healthcare has rapidly adopted EHR systems with embedded AI for data capture and population; major hospital systems and oncology centers are actively deploying automated record-keeping workflows.
Sector adoption velocity	claude-sonnet-5	3/5	Radiation oncology has moderate digitization with treatment planning and record-and-verify systems widely used, but broader AI-driven documentation adoption is still emerging compared to faster-moving sectors like finance.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI systems substantially assist dosimetrists by auto-populating records from treatment plans and imaging systems, reducing manual entry burden and enabling the human to focus on verification and complex clinical decisions.
Augmentation potential	claude-sonnet-5	4/5	AI and automated systems can significantly reduce manual transcription effort and error by auto-populating dose records from treatment planning outputs, letting dosimetrists focus on verification.
Task automatability	claude-haiku-4-5-20251001	4/5	Recording structured patient information like radiation doses is highly routine data entry that current AI systems can reliably extract from source documents and populate into electronic records with minimal human review, meeting the 50% time-saving threshold at comparable quality.
Task automatability	claude-sonnet-5	4/5	Recording structured dose data into records is a well-defined data entry/transcription task that can largely be automated via integration between treatment planning systems and EHR/record-and-verify systems, with AI/automation handling extraction and logging.
Adoption barriers	claude-haiku-4-5-20251001	3/5	While data entry itself has low barriers, healthcare regulations (HIPAA, audit trails) and the requirement that licensed dosimetrists or physicians verify dosimetry data before clinical use create moderate friction to full automation.
Adoption barriers	claude-sonnet-5	3/5	Accurate radiation dose records are medico-legally significant and subject to regulatory documentation standards, requiring human verification/sign-off even if data capture is automated.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	The cost of AI-driven data entry and record management is orders of magnitude cheaper than human clerical time once integrated into existing EHR workflows.
Cost vs. human wage	claude-sonnet-5	4/5	Automated data logging via integrated software is far cheaper than manual dosimetrist time spent transcribing values, though initial integration and validation costs exist.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	EHR systems with AI-assisted data entry and auto-population from structured sources are deployed in many healthcare settings; OCR and information extraction for medical documents are mature technologies, though integration depth varies by institution.
Technical feasibility today	claude-sonnet-5	3/5	Record-and-verify systems and EHR integrations already automate much of this data transfer in production oncology departments, though manual verification and entry of some fields still occurs, and full end-to-end AI-driven documentation is not universal.

Calculate, or verify calculations of, prescribed radiation doses.

CI 34–57 · exposure 58 · augmentation 75 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Medical dosimetry software is in routine use in hospitals and cancer centers, but adoption of fully autonomous AI verification remains in pilot/limited production phases; the sector digitizes steadily but regulatory caution moderates rapid displacement.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare/radiation oncology is a highly regulated, safety-critical sector with slow, cautious technology adoption cycles, especially for tasks involving direct patient dose determination.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-assisted dose calculation and verification tools can substantially raise dosimetrist productivity by automating routine computations and flagging inconsistencies, allowing the human expert to focus on complex planning and clinical judgment.
Augmentation potential	claude-sonnet-5	4/5	AI-assisted dose calculation engines and automated secondary checks already meaningfully speed up and improve accuracy of the calculation workflow while the dosimetrist remains responsible for final verification.
Task automatability	claude-haiku-4-5-20251001	4/5	Current AI can perform dose calculations with high accuracy using algorithms and verify numerical outputs reliably; however, clinical integration and quality assurance oversight typically still require human review, preventing full end-to-end automation without human involvement in validation steps.
Task automatability	claude-sonnet-5	3/5	Dose calculations are structured numerical problems that treatment planning systems already automate substantially, but the prescribed dose verification step requires clinical judgment and safety-critical checking that current AI cannot fully replace end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory barriers are substantial: FDA oversight of dosimetry software, state licensure requirements for dosimetrists, and liability and error-cost asymmetry in radiation oncology create meaningful friction preventing full substitution without licensed human sign-off.
Adoption barriers	claude-sonnet-5	5/5	Radiation dose verification is heavily regulated (e.g., NRC, state licensing, accreditation standards) and legally requires sign-off by credentialed medical physicists/dosimetrists, making autonomous AI substitution essentially prohibited.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI-driven dose calculation and verification incurs primarily software licensing and integration costs, which are substantially lower than the loaded wage of a medical dosimetrist performing these tasks manually.
Cost vs. human wage	claude-sonnet-5	2/5	TPS software licensing, calibration, and mandatory human verification overhead keep costs comparable to or only modestly below the loaded cost of a trained dosimetrist, since redundant human checks remain required for patient safety.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Dose calculation software exists and is deployed in clinical settings, but these are typically narrow-scope tools requiring expert configuration and human verification rather than fully autonomous end-to-end systems; AI verification of calculations is emerging but not yet mature in production at scale.
Technical feasibility today	claude-sonnet-5	3/5	Commercial treatment planning software (TPS) and secondary dose-check tools (e.g., Monte Carlo verification systems) are deployed in clinics today, but they function as decision-support requiring dosimetrist/physicist sign-off rather than autonomous calculation and verification.

Fabricate patient immobilization devices, such as molds or casts, for radiation delivery.

CI 5–82 · exposure 45 · augmentation 50 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Large medical centers and cancer treatment facilities are actively adopting 3D printing and automated fabrication for patient-specific devices; this is a relatively digitized, high-value healthcare application with measurable pilot-to-production adoption in the last 3–5 years, though smaller facilities lag.
Sector adoption velocity	claude-sonnet-5	1/5	Radiation oncology fabrication workflows remain physically manual and have seen negligible AI/robotic adoption for immobilization device creation.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI and automated fabrication significantly augment dosimetrists' productivity by removing the manual molding/casting bottleneck, enabling faster turnaround, better precision, and fewer remakes—while the dosimetrist remains responsible for quality verification and clinical validation of the output.
Augmentation potential	claude-sonnet-5	2/5	AI/3D printing design software can assist in designing custom mold geometries or optimizing fit parameters, but the physical fabrication and fitting process itself sees minimal AI augmentation.
Task automatability	claude-haiku-4-5-20251001	5/5	Fabricating immobilization devices (molds, casts) for radiation is a highly structured manufacturing task involving 3D scanning, CAD modeling, and precision fabrication—processes where current AI and automation systems (3D printing, CNC milling, robotic arms) achieve >50% time savings at equal or better quality. End-to-end automation from patient anatomy to finished device is demonstrably feasible with existing technology.
Task automatability	claude-sonnet-5	1/5	This is a hands-on physical fabrication task requiring manual molding, fitting to a patient's body, and material handling that current AI systems cannot perform without robotic embodiment.
Adoption barriers	claude-haiku-4-5-20251001	3/5	While the fabrication itself has minimal regulatory barriers, clinical use of custom immobilization devices requires quality assurance, physician oversight, and validation that the final device meets safety specs—creating oversight friction. No single-license requirement prevents automation, but integration into validated clinical workflows introduces organizational friction.
Adoption barriers	claude-sonnet-5	4/5	Direct patient contact, precise anatomical fitting, and clinical safety requirements for radiation therapy accessories create strong practical and quality-control barriers, though not necessarily a strict licensure requirement for this specific sub-task.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated 3D printing and robotic fabrication of custom immobilization devices cost a small fraction of the labor required for manual molding and casting by skilled technicians, easily achieving an order-of-magnitude cost reduction once equipment is in place.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI-driven substitute for physical mold/cast fabrication, so any AI-based approach would require costly robotic hardware exceeding current human labor costs.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Production AI systems and industrial automation (3D printers, robotic fabrication) already perform custom immobilization device manufacturing in clinical and research settings, though integration with clinical workflows and QA requirements introduces some friction. The core fabrication task is mature and deployed, though clinical integration remains partly manual.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI or robotic product fabricates patient-specific immobilization devices in clinical radiation oncology settings today; this remains a manual technologist/dosimetrist task.

Perform quality assurance system checks, such as calibrations, on treatment planning computers.

CI 0–78 · exposure 45 · augmentation 63 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Hospital radiation oncology departments and large treatment centers have already adopted automated QA monitoring systems as standard practice; uptake is strong in digitized medical settings, though smaller or under-resourced facilities may lag.
Sector adoption velocity	claude-sonnet-5	1/5	Radiation oncology physics QA is a highly specialized, safety-critical, low-digitization niche with minimal AI agent deployment for physical calibration tasks.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-based QA systems meaningfully augment dosimetrist productivity by automating routine checks, flagging anomalies, and generating reports, allowing dosimetrists to focus on interpretation, corrective action, and complex problem-solving rather than manual measurement.
Augmentation potential	claude-sonnet-5	3/5	AI-based software can flag anomalies, trend calibration data, or assist in scheduling/documentation of QA checks, providing moderate assistance while humans perform the actual verification.
Task automatability	claude-haiku-4-5-20251001	5/5	AI-driven systems can execute routine calibration checks, data validation, and anomaly detection on treatment planning computers with minimal human intervention, easily meeting the 50% time-saving threshold. The task involves standardized, repeatable procedures (comparing measured values to expected ranges, logging results) that current automated quality assurance software handles reliably.
Task automatability	claude-sonnet-5	1/5	This requires hands-on physical calibration checks and verification against radiation delivery hardware, which AI cannot physically perform or independently validate today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Medical device regulations (FDA, IEC standards for treatment planning systems) require documented QA procedures and often mandate human review/sign-off of critical calibration results, creating a regulatory requirement for human oversight that limits full automation.
Adoption barriers	claude-sonnet-5	5/5	Radiation therapy QA is heavily regulated (e.g., accreditation bodies, state/federal radiation safety rules) and typically requires certified/licensed medical physicists or dosimetrists to sign off, creating hard legal barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated QA systems run continuously at negligible marginal cost per check compared to the loaded labor cost of a dosimetrist performing manual calibration verification, achieving an order-of-magnitude cost advantage.
Cost vs. human wage	claude-sonnet-5	1/5	Since AI cannot perform the physical calibration and verification steps, there is no viable AI substitute cost to compare against the human's loaded wage.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature automated QA systems exist and are deployed in clinical settings for routine calibration monitoring and system checks; however, complex hardware failures or edge cases may still require expert interpretation, preventing a full 5 rating.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product autonomously executes treatment planning computer QA calibration checks in clinical practice; this remains a manual, regulated physicist/dosimetrist task.

Identify and outline bodily structures, using imaging procedures, such as x-ray, magnetic resonance imaging, computed tomography, or positron emission tomography.

CI 30–45 · exposure 42 · augmentation 75 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption remains slow and cautious in radiology and oncology practices. While pilots of automated segmentation are underway, most departments still rely heavily on manual contouring due to liability concerns, regulatory caution, and the critical nature of dosimetry planning. Institutional inertia and reimbursement structures limit rapid uptake.
Sector adoption velocity	claude-sonnet-5	3/5	Radiation oncology departments are progressively adopting AI auto-contouring tools, but adoption is uneven across institutions and still requires human verification, reflecting a healthcare sector with moderate digitization and cautious rollout.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI segmentation tools significantly augment dosimetrist productivity by providing automated initial contours that can be refined rather than drawn from scratch, substantially accelerating the workflow. Dosimetrists remain in the loop for quality control and clinical judgment, making this a strong assistive application.
Augmentation potential	claude-sonnet-5	4/5	AI auto-segmentation substantially speeds up the initial contouring step, letting dosimetrists focus on refinement and quality assurance rather than manual delineation from scratch.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can assist with segmentation of anatomical structures from imaging, but clinical dosimetry requires precise manual contouring for treatment planning with critical safety implications. Current automated segmentation tools have meaningful error rates that necessitate substantial human review and manual correction, falling short of the 50% time-savings-at-equal-quality threshold.
Task automatability	claude-sonnet-5	3/5	AI-based auto-contouring tools can segment many normal organs-at-risk from CT/MRI with reasonable accuracy, saving time, but tumor volumes and complex structures still require substantial expert correction, so full end-to-end automation at equal quality is not yet achieved.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong regulatory barriers exist: FDA oversight of software as a medical device, liability for dosimetry errors (which directly affect patient safety), and clinical practice standards that typically require a licensed dosimetrist or radiologist to validate contours before treatment. The high error-cost asymmetry and human sign-off requirement create substantial friction.
Adoption barriers	claude-sonnet-5	4/5	Radiation treatment planning is a regulated clinical process requiring qualified medical dosimetrists/physicians to verify and approve contours, given high liability for treatment errors, creating strong sign-off barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI segmentation software and integration costs are substantial, and human oversight remains mandatory for quality assurance and liability reasons. The human dosimetrist time saved is offset by licensing, infrastructure, and the need for expert oversight, making costs roughly comparable to or exceeding human-only workflows.
Cost vs. human wage	claude-sonnet-5	3/5	Auto-contouring software has licensing and computation costs but reduces dosimetrist time per case significantly; net savings exist but are moderate once integration, QA, and correction time are factored in.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Semi-automated segmentation products exist in clinical workflows (e.g., atlas-based and deep learning tools), but they require significant manual refinement by dosimetrists and radiologists before clinical use. These systems perform specific structures with variable accuracy and are rarely fully autonomous in production.
Technical feasibility today	claude-sonnet-5	3/5	Deployed auto-segmentation products (e.g., MIM, RayStation, Limbus AI) are used clinically for organ-at-risk contouring, but dosimetrists still routinely review and edit outputs due to error rates, especially near tumor boundaries or atypical anatomy.

Calculate the delivery of radiation treatment, such as the amount or extent of radiation per session, based on the prescribed course of radiation therapy.

CI 29–39 · exposure 45 · augmentation 88 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Oncology centers adopt AI-assisted planning tools slowly due to regulatory requirements, training burdens, and the need for clinical validation before deployment. Adoption remains in pilot and early-production phases in major academic centers rather than widespread deployment across the sector.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially clinical radiation oncology, is a slow-adopting sector for full automation due to regulatory oversight and patient safety concerns, though computer-assisted planning tools have been standard for years.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI dose-optimization and planning assistance can significantly boost dosimetrist productivity by automating iterative calculation, suggesting plan refinements, and flagging anomalies. This allows dosimetrists to focus on clinical judgment and complex cases while AI handles computational heavy lifting and routine validation checks.
Augmentation potential	claude-sonnet-5	5/5	AI-driven dose optimization and knowledge-based planning tools substantially speed up and improve consistency of dose calculations while the dosimetrist remains responsible for review and adjustment.
Task automatability	claude-haiku-4-5-20251001	3/5	Dose calculation involves complex mathematical modeling of radiation physics that current AI systems can support, but integration with medical imaging, patient anatomy interpretation, and clinical decision-making requires significant human oversight and setup. AI can automate parts of the computation, but not the full end-to-end task including validation and clinical judgment.
Task automatability	claude-sonnet-5	3/5	Treatment planning systems already automate much of the dose calculation via optimization algorithms, but final calculations require clinical judgment, plan verification, and patient-specific adaptation that AI cannot fully own end-to-end today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory frameworks (FDA, state medical boards) require a licensed medical dosimetrist or radiation oncologist to review and approve dose calculations and treatment plans. Liability concerns around radiation safety and patient harm create strong legal and organizational barriers to full automation without human sign-off.
Adoption barriers	claude-sonnet-5	5/5	Radiation dose calculations are safety-critical and heavily regulated; certified medical dosimetrists/physicists must review and sign off on all treatment plans, making full automation legally and clinically prohibited.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Implementation of AI dose-calculation systems requires specialized clinical software, integration with oncology planning systems, and ongoing oversight by trained dosimetrists. The combined cost of the system, training, validation, and required human review remains comparable to or exceeds the cost of direct dosimetrist labor.
Cost vs. human wage	claude-sonnet-5	2/5	Treatment planning software is expensive to license and requires substantial IT/clinical integration and mandatory human oversight, so total cost savings versus a trained dosimetrist are moderate, not order-of-magnitude.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While research systems and some clinical software include AI-assisted dose optimization, deployed products in real hospital workflows still require dosimetrists to review, validate, and refine calculations. Material error rates and the need for human sign-off prevent full automation in production settings.
Technical feasibility today	claude-sonnet-5	3/5	Commercial treatment planning software (e.g., automated planning modules, knowledge-based planning) is deployed in clinics, but dosimetrists still manually verify, adjust, and finalize calculations due to safety-critical error tolerances.

Design the arrangement of radiation fields to reduce exposure to critical patient structures, such as organs, using computers, manuals, and guides.

CI 30–32 · exposure 34 · augmentation 88 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare automation is slower than information-sector adoption due to regulatory oversight, liability concerns, and the need for clinical validation. While treatment planning software is widely deployed, meaningful displacement of dosimetrist labor through AI agents has not materialized; adoption remains at the tool-assistance level.
Sector adoption velocity	claude-sonnet-5	3/5	Radiation oncology has adopted knowledge-based planning and AI-assisted optimization tools at a moderate pace, with pilots and partial integration common in academic and larger centers, but not yet universal or fully autonomous.
Augmentation potential	claude-haiku-4-5-20251001	4/5	Modern treatment planning software (dose engines, optimization algorithms, constraint-based planning tools) substantially amplifies a dosimetrist's productivity by accelerating iteration and exploring more candidate field arrangements than manual design would allow. AI-driven dose optimization and anatomy contouring suggestions directly enhance the human's ability to refine plans.
Augmentation potential	claude-sonnet-5	5/5	AI-driven automated planning and optimization tools significantly speed up initial plan generation and iteration, allowing dosimetrists to focus on fine-tuning and clinical judgment, substantially boosting productivity.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with dose optimization algorithms and field arrangement suggestions, the task requires real-time clinical judgment to balance competing priorities (tumor coverage vs. organ protection) and adapt to individual patient anatomy in ways current systems cannot fully automate end-to-end. Human dosimetrists remain essential for validation and final optimization.
Task automatability	claude-sonnet-5	2/5	Radiation field/beam arrangement optimization involves complex 3D anatomical reasoning and treatment planning software (TPS) already has optimization algorithms, but final design still requires substantial clinical judgment and iterative human refinement, so full end-to-end automation at equal quality is not yet achieved.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory bodies (FDA, ACR) mandate that a licensed Medical Dosimetrist or Radiation Oncologist must sign off on final treatment plans, and liability for dose errors falls on the clinic and supervising physician. These legal and professional accountability requirements create substantial barriers to full autonomy.
Adoption barriers	claude-sonnet-5	4/5	Treatment plans require sign-off by certified medical dosimetrists and radiation oncologists/physicists due to patient safety and regulatory requirements (e.g., FDA, accreditation bodies), creating strong barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Computational dose optimization is relatively cheap per run, but the total cost including clinical validation, oversight, and integration into existing workflows remains high. A dosimetrist's labor still dominates the cost per completed plan; automation saves perhaps 20–30% of time, not enough for dramatic cost advantage.
Cost vs. human wage	claude-sonnet-5	2/5	TPS software and optimization algorithms have high licensing/maintenance costs and still require a trained dosimetrist's oversight and adjustment time, so total cost is not dramatically lower than employing a dosimetrist directly.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Treatment planning systems with dose calculation and optimization modules exist in clinical use, but they function as tools requiring significant human oversight rather than autonomous agents. Systems like inverse planning algorithms assist but do not independently design field arrangements meeting the full clinical standard; human dosimetrists must review, validate, and refine all proposals.
Technical feasibility today	claude-sonnet-5	3/5	Commercial treatment planning systems (e.g., Eclipse, RayStation) include automated/knowledge-based planning (KBP) and inverse optimization tools deployed in clinics, but dosimetrists still manually adjust and validate plans, so it's assistive rather than fully autonomous.

Educate patients regarding treatment plans, physiological reactions to treatment, or post-treatment care.

CI 25–25 · exposure 25 · augmentation 75 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare adoption of AI for patient-facing clinical education remains slow due to liability concerns, regulatory caution, and institutional preference for human accountability. While some health systems pilot chatbots for basic triage or appointments, specialized clinical education by AI remains rare in production.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially radiation oncology, is a highly regulated, lower-digitization sector with cautious, slow AI adoption for direct patient-facing communication tasks.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI could significantly assist dosimetrists by drafting patient-friendly explanations, generating visual aids, or preparing personalized educational materials that the dosimetrist then delivers and refines. This keeps the human expert in the loop while automating content generation and research.
Augmentation potential	claude-sonnet-5	4/5	AI can generate patient-friendly explanations, translate materials, and provide draft FAQs that dosimetrists or clinicians can review and personalize, meaningfully speeding up preparation of patient education content.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI could generate educational content or draft explanations of treatment plans, the task requires personalized communication tailored to individual patient concerns, comprehension levels, and emotional state. Current AI systems lack the ability to reliably adapt messaging in real-time based on patient feedback and cannot replace the human judgment needed to assess whether a patient truly understands.
Task automatability	claude-sonnet-5	2/5	Explaining general information could be drafted by AI, but personalized, empathetic patient education requires real-time interaction, reading emotional cues, and answering unpredictable clinical questions tied to a specific treatment plan.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Patient education in a clinical context carries legal and liability implications; most healthcare systems require a licensed professional (dosimetrist or physician) to verify and take responsibility for patient information accuracy. Patients also often prefer and expect human interaction for sensitive health discussions, and regulatory frameworks place accountability on the organization for patient comprehension.
Adoption barriers	claude-sonnet-5	4/5	Patient education about treatment risks and care is often legally and ethically tied to licensed clinical staff, with liability concerns and regulatory expectations around informed consent limiting full AI substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI infrastructure for reliable patient education requires specialized fine-tuning, human oversight for safety, and integration with clinical workflows. When accounting for validation, liability oversight, and the labor cost of training and monitoring the system, the total cost approaches or exceeds that of a dosimetrist providing education.
Cost vs. human wage	claude-sonnet-5	2/5	While AI content generation is cheap, the need for clinician verification, liability review, and in-person reassurance keeps effective all-in costs comparable to or only modestly below human-delivered education.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed clinical product reliably handles personalized patient education on treatment physiological reactions independently. AI chatbots exist for basic health information, but they lack the specialized dosimetry knowledge, real-time assessment of patient understanding, and accountability required in clinical settings.
Technical feasibility today	claude-sonnet-5	2/5	Chatbots and AI patient-education tools exist for general oncology information, but no deployed product reliably conducts personalized clinical dosimetry education with the accuracy and trust required in production clinical workflows.

Develop radiation treatment plans in consultation with members of the radiation oncology team.

CI 20–28 · exposure 25 · augmentation 63 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Oncology and medical physics remain relatively conservative sectors with slow digital transformation. While planning-support tools are emerging in academic centers, production deployment of autonomous AI plan development is minimal and adoption remains in pilot phases.
Sector adoption velocity	claude-sonnet-5	3/5	Radiation oncology departments have adopted AI-assisted planning tools (auto-contouring, knowledge-based planning) at a moderate pace, with pilots and partial deployment common but full automation rare due to safety-critical nature.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist dosimetrists by automating dose calculations, suggesting plan optimizations, and flagging anatomical landmarks, thereby raising planning speed and consistency. However, the human dosimetrist remains essential for final clinical judgment and regulatory compliance.
Augmentation potential	claude-sonnet-5	4/5	AI substantially augments dosimetrists by auto-generating initial dose distributions, optimizing plans faster, and flagging suboptimal areas, while the dosimetrist and team retain final clinical judgment and consultation.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist in dose calculations and plan optimization, developing treatment plans requires complex clinical judgment, patient-specific anatomical interpretation, and consultation with multidisciplinary teams. Current AI systems lack the contextual reasoning and collaborative capability to replace this end-to-end at the required quality threshold.
Task automatability	claude-sonnet-5	2/5	AI can assist in generating candidate dose distributions and plan optimization, but the task as stated is a collaborative clinical decision requiring judgment across the oncology team that current systems cannot fully replace end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory bodies (FDA, state medical boards) require licensed medical dosimetrists or physicians to sign off on radiation treatment plans; automation of the entire task faces legal and liability barriers. Patient safety and error-cost asymmetry create strong organizational and regulatory friction.
Adoption barriers	claude-sonnet-5	5/5	Radiation treatment plans require sign-off by licensed medical physicists and radiation oncologists under strict regulatory and safety frameworks (e.g., FDA, ASTRO, accreditation bodies), making full automation legally and clinically prohibited.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-assisted planning tools require substantial infrastructure, integration with existing clinical systems, and expert validation overhead. The loaded cost of a dosimetrist remains competitive with or lower than the total cost of implementing and maintaining AI systems that still require significant human review.
Cost vs. human wage	claude-sonnet-5	2/5	AI-assisted planning software reduces some planning time but requires expensive TPS licensing, integration with clinical workflows, and mandatory human oversight, keeping all-in costs comparable to or only modestly below human-only planning.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Research systems exist for dose prediction and plan optimization, but no deployed product reliably performs independent treatment plan development in clinical production. Clinical adoption remains limited to narrow, well-constrained sub-tasks (e.g., DVH prediction) with significant human oversight.
Technical feasibility today	claude-sonnet-5	2/5	Automated planning tools (e.g., knowledge-based planning, deep-learning dose prediction) exist in commercial treatment planning systems, but they generate draft plans requiring dosimetrist/physicist review and team consultation, not autonomous plan finalization.

Teach medical dosimetry, including its application, to students, radiation therapists, or residents.

CI 21–25 · exposure 25 · augmentation 63 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Medical education and clinical training remain highly conservative and human-centered; while online learning tools are slowly adopted, teaching of technical medical skills relies on apprenticeship with licensed professionals. Substitution of AI for instructors is extremely limited even in the most digitized healthcare organizations.
Sector adoption velocity	claude-sonnet-5	2/5	Medical/clinical education is a highly regulated, in-person-focused field with slow AI adoption for core teaching and certification functions.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by generating course slides, providing supplementary worked examples, quizzes, or instant reference answers to routine questions, thereby raising an instructor's preparation efficiency. However, the core act of teaching—live explanation, assessment, and mentoring—remains primarily human, limiting augmentation to administrative and content-support functions.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by generating lecture materials, practice case studies, quiz questions, and explanatory content, augmenting instructor efficiency significantly while humans remain the primary teachers.
Task automatability	claude-haiku-4-5-20251001	2/5	Teaching involves complex pedagogical judgment, adapting explanations to student comprehension, and live feedback that current AI systems cannot fully replicate reliably. While AI can draft course materials or explain concepts, it cannot manage a classroom interaction, assess individual learning needs in real time, or adjust instructional pace—core components of effective teaching.
Task automatability	claude-sonnet-5	2/5	Teaching involves live demonstration, clinical case walkthroughs, and responsive mentorship that current AI cannot fully replicate end-to-end despite generating some instructional content.'},
Adoption barriers	claude-haiku-4-5-20251001	4/5	Teaching in a regulated medical field carries strong institutional, liability, and accreditation barriers; regulatory bodies (AAPM, ACR) expect human instructors to be qualified and licensed, and medical education requires face-to-face clinical mentorship by law or professional standard. An AI cannot be held accountable for student competency or licensing outcomes.
Adoption barriers	claude-sonnet-5	4/5	Clinical education programs typically require credentialed, experienced dosimetrists or radiation therapists to teach and evaluate trainees per accreditation standards, creating strong barriers to full AI substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	A qualified medical dosimetrist or physicist commanding a professional wage is far cheaper per student when delivering instruction than the combined cost of developing, maintaining, and deploying AI with sufficient oversight, human review, and clinical validation for medical education.
Cost vs. human wage	claude-sonnet-5	2/5	Human instructors combine clinical judgment, hands-on supervision, and accreditation-required teaching that AI cannot cheaply replace without significant human oversight still needed.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably teaches medical dosimetry end-to-end; AI can assist with content generation and tutoring modules, but these lack the nuanced mentorship, clinical context-setting, and real-time adaptation that professional teaching in medical physics requires. Existing educational AI tools are narrow and don't substitute for instructor-led clinical training.
Technical feasibility today	claude-sonnet-5	2/5	Some AI tutoring and content-generation tools exist but no deployed product reliably teaches clinical dosimetry practice to students or residents in real training programs.

Develop requirements for the use of patient immobilization devices and positioning aides, such as molds or casts, as part of treatment plans to ensure accurate delivery of radiation and comfort of patient.

CI 16–25 · exposure 20 · augmentation 50 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare and radiotherapy centers adopt new technologies cautiously; pilot programs for automated immobilization planning exist but are rare in production. The sector is digitizing slowly compared to IT/finance, and adoption remains limited to research centers and large specialized facilities.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare/radiation oncology is a highly regulated, moderately slow-adopting sector for autonomous clinical decisions, though AI tools are increasingly used for treatment planning support rather than physical setup decisions.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist dosimetrists by analyzing imaging, suggesting device types and positioning parameters, and flagging anatomical constraints—useful productivity aids. However, the human dosimetrist must validate clinical appropriateness, comfort trade-offs, and patient-specific factors, keeping augmentation at moderate strength.
Augmentation potential	claude-sonnet-5	3/5	AI can help by suggesting standard immobilization protocols based on treatment site/history or flagging setup inconsistencies, providing moderate assistance while the dosimetrist retains responsibility for the physical and clinical decision.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can analyze imaging and suggest immobilization parameters, the task requires clinical judgment about patient-specific anatomy, comfort, and safety—factors that demand human oversight. Current systems can assist with requirements generation but cannot reliably substitute for the full task end-to-end with consistent quality.
Task automatability	claude-sonnet-5	2/5	This requires physical assessment of patient anatomy, comfort, and setup reproducibility, which involves hands-on judgment and physical fitting that current AI cannot perform end-to-end; AI can assist with documentation or protocol lookup but not the core physical/clinical decision-making.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Medical device development and patient safety requirements create significant regulatory and liability barriers; immobilization directly affects radiation accuracy and patient safety. Clinical judgment and sign-off by a qualified dosimetrist are typically required by institutional protocols and regulatory standards.
Adoption barriers	claude-sonnet-5	4/5	Clinical radiation oncology workflows require credentialed staff to specify immobilization and positioning for patient safety and treatment accuracy, with significant liability and regulatory oversight (e.g., accreditation, physics review) tied to this decision.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for immobilization planning are expensive to acquire and integrate into clinical workflows, while dosimetrists' labor is already embedded in treatment centers. The all-in cost of AI deployment (software, validation, training) currently exceeds the cost of human dosimetrists performing this task.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute performing this physical-and-judgment task, so the all-in AI cost for full task completion is effectively infeasible compared to human labor.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No mature production systems currently perform this task autonomously. Some research applications exist for automated immobilization planning from imaging data, but deployed clinical tools still rely on dosimetrists' expertise to finalize device specifications and validate patient comfort.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously specifies immobilization device requirements or positioning aids in clinical practice; this remains a human dosimetrist/therapist task requiring physical patient interaction.

Plan the use of beam modifying devices, such as compensators, shields, and wedge filters, to ensure safe and effective delivery of radiation treatment.

CI 20–20 · exposure 25 · augmentation 75 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	While large cancer centers use advanced treatment planning software, adoption of autonomous beam-configuration agents remains minimal; most clinics still rely on human dosimetrists to drive planning decisions, with AI serving a supporting rather than leading role.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially radiation oncology, is a highly regulated and cautious sector with slow, incremental adoption of AI tools embedded within existing planning software rather than fast deep transformation.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI systems routinely assist dosimetrists by rapidly generating initial beam configurations, optimizing dose distributions, and flagging constraint violations, substantially speeding up plan creation and refinement while the dosimetrist retains final clinical judgment.
Augmentation potential	claude-sonnet-5	4/5	AI-based optimization algorithms in treatment planning systems meaningfully speed up generation of candidate beam arrangements and modifier configurations, letting dosimetrists focus on review, adjustment, and safety verification.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist in calculating beam parameters and suggesting shielding configurations, the task requires integration of patient-specific anatomy, clinical judgment about trade-offs between dose conformity and treatment time, and real-time decision-making that current systems cannot fully automate end-to-end without substantial human oversight.
Task automatability	claude-sonnet-5	2/5	Treatment planning systems can optimize beam shaping with algorithms, but final selection of compensators, shields, and wedges requires clinical judgment, patient-specific anatomy review, and safety verification that current AI cannot fully replace end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Medical physics and dosimetry planning are subject to strict regulatory oversight (FDA, state medical boards), clinical protocol requirements, and liability law that typically mandates a licensed medical dosimetrist or physicist to review, validate, and take responsibility for the treatment plan before clinical use.
Adoption barriers	claude-sonnet-5	5/5	Radiation treatment planning is heavily regulated, requires certified/licensed medical dosimetrists or physicists, and errors carry severe patient safety and liability consequences, mandating human sign-off by law and clinical protocol.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current AI planning tools are expensive licensing solutions integrated into clinic workflows, and they still require skilled dosimetrist time for review and modification, making total cost per treatment plan comparable to or higher than human-only planning.
Cost vs. human wage	claude-sonnet-5	2/5	Specialized treatment planning software and computation infrastructure carry significant licensing and integration costs, and mandatory human oversight for safety-critical radiation dosing keeps the effective cost comparable to or only modestly below skilled human labor.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Treatment planning software exists and incorporates dose calculation, but deployed systems function as decision-support tools requiring dosimetrist expertise to validate, adjust, and approve configurations; no current product reliably performs this task independently at clinical standard.
Technical feasibility today	claude-sonnet-5	2/5	Some treatment planning software includes automated optimization features, but these are decision-support tools requiring dosimetrist review and adjustment rather than autonomous, reliable production systems for this specific task.

Create and transfer reference images and localization markers for treatment delivery, using image-guided radiation therapy.

CI 20–20 · exposure 25 · augmentation 63 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption of fully autonomous AI for reference image creation and marker transfer in radiation oncology remains slow, with most facilities still using human dosimetrists as the primary agent; pilot projects exist but production displacement is minimal due to regulatory caution and quality assurance requirements.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare/radiation oncology adopts new imaging and AI tools cautiously due to regulatory review, safety validation, and reimbursement hurdles, resulting in slow, incremental AI integration in this specific niche.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist dosimetrists by automating image registration, suggesting marker positions, and flagging potential alignment issues, thereby reducing manual effort on parts of the task while the dosimetrist retains control over final validation and clinical decisions.
Augmentation potential	claude-sonnet-5	4/5	AI-based image registration, contouring, and localization tools significantly speed up marker placement and image preparation, letting dosimetrists focus on verification and complex cases while staying in the loop.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with image analysis and marker detection, the task requires precise anatomical localization, spatial alignment decisions, and clinical judgment that demand human oversight. Current AI systems lack the reliability and adaptability to handle the full end-to-end workflow of creating and transferring reference images with the consistency and quality assurance standards required in radiation therapy.
Task automatability	claude-sonnet-5	2/5	Involves hands-on integration with imaging and treatment delivery hardware plus clinical judgment about localization accuracy, which current AI cannot fully perform end-to-end. Some image processing steps could be assisted but the full workflow requires physical/system interaction and verification.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Radiation therapy is heavily regulated by medical boards and physics-based licensing requirements; image localization and treatment guidance directly impact patient safety and dose delivery, creating legal and liability barriers that require licensed professionals to validate and sign off on marker placement and image transfers.
Adoption barriers	claude-sonnet-5	5/5	This task is embedded in a regulated clinical radiation therapy workflow requiring certified medical dosimetrists/physicists to verify accuracy before treatment, with high liability for errors—strong legal and safety barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current AI-assisted tools require significant infrastructure, specialized software integration, and human oversight that approaches or exceeds the cost-benefit of having trained dosimetrists perform the task directly, especially given liability and quality assurance demands.
Cost vs. human wage	claude-sonnet-5	2/5	Specialized medical imaging AI tools carry high validation, integration, and regulatory costs, and clinical oversight is mandatory, keeping cost savings modest compared to human specialist time.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Some image analysis and registration tools exist in clinical workflows, but fully autonomous creation and transfer of reference images with localization markers remains limited to research or early-stage implementations. Deployed systems are typically narrow in scope and still require substantial human verification and adjustment.
Technical feasibility today	claude-sonnet-5	2/5	While image-guided radiation therapy (IGRT) software includes automated registration and marker detection features, transferring these to treatment delivery reliably still requires dosimetrist oversight and manual verification in production clinical settings.

Develop treatment plans, and calculate doses for brachytherapy procedures.

CI 16–20 · exposure 25 · augmentation 63 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare AI adoption in clinical physics remains cautious and heavily regulated; brachytherapy is a specialized, lower-volume procedure. Adoption of AI assistance in dose calculation exists in research and pilot settings but has not achieved deep production-level displacement in routine clinical workflows.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare and radiation oncology are historically slow to adopt autonomous AI systems for safety-critical dosing decisions, with adoption limited mostly to decision-support tools rather than full automation.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist dosimetrists by automating routine dose calculations, suggesting optimized plans, and flagging outliers, raising their efficiency on plan generation and validation. However, the core task of clinical judgment and plan approval remains human-centered, limiting transformative augmentation.
Augmentation potential	claude-sonnet-5	4/5	AI-driven dose optimization algorithms and automated planning tools meaningfully speed up plan generation and allow dosimetrists to explore more options, significantly boosting productivity while the human retains final responsibility.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with dose calculations and reference treatment protocols, the task requires integration of patient anatomy, physics, clinical judgment, and regulatory compliance that current systems cannot fully automate end-to-end. Dosimetrists must validate and sign off on plans, and significant clinical oversight remains mandatory.
Task automatability	claude-sonnet-5	2/5	Treatment planning optimization software already assists with dose calculations, but the full task requires integrating patient anatomy, clinical judgment, and safety verification that current AI cannot autonomously complete end-to-end at equal quality without significant human oversight.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Medical dosimetry is a licensed/certified profession in many jurisdictions, and brachytherapy dose calculation and treatment plan approval carry high liability and regulatory oversight. Regulatory bodies (FDA, state medical boards) require a qualified human professional to develop and certify treatment plans.
Adoption barriers	claude-sonnet-5	5/5	Brachytherapy dose planning is tightly regulated by medical physics and radiation oncology licensing bodies, with mandatory physician/physicist sign-off due to high liability from radiation errors, making unsupervised AI substitution legally prohibited.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	Brachytherapy treatment planning requires specialized software, physics expertise, and regulatory compliance; the all-in cost of AI systems with necessary oversight, validation, and human expertise integration remains higher than direct dosimetrist labor for this safety-critical task.
Cost vs. human wage	claude-sonnet-5	2/5	Specialized medical physics software and computational modeling require significant licensing, validation, and integration costs, and human oversight remains mandatory, keeping costs comparable to or only modestly below human labor costs.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Some commercial treatment planning systems incorporate AI-assisted modules for dose optimization and planning suggestions, but no deployed product performs the complete brachytherapy treatment plan development and dose calculation without human dosimetrist review and approval. Regulatory requirements mandate human professional sign-off.
Technical feasibility today	claude-sonnet-5	2/5	Commercial treatment planning systems include algorithmic dose optimization tools, but no deployed product independently generates and validates full brachytherapy plans without a certified dosimetrist's active involvement.

Supervise or perform simulations for tumor localizations, using imaging methods such as magnetic resonance imaging, computed tomography, or positron emission tomography scans.

CI 3–25 · exposure 13 · augmentation 63 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare remains a laggard sector for autonomous AI deployment due to regulatory oversight, liability concerns, and institutional conservatism; while AI-assisted analysis tools are adopted for speed, independent tumor localization automation is rare in production.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare imaging workflows are adopting AI tools slowly for image analysis assistance, but the physical simulation task itself remains largely untouched by automation in production settings.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-driven image segmentation, registration, and anatomical localization tools significantly augment dosimetrist productivity by automating routine contour suggestions and pre-processing while the human dosimetrist retains clinical oversight and refinement of the simulation.
Augmentation potential	claude-sonnet-5	3/5	AI-based image segmentation and contouring tools can assist dosimetrists in identifying tumor boundaries and structures from CT/MRI/PET scans, improving efficiency during the planning phase.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist in image analysis and segmentation from MRI, CT, and PET scans, the full end-to-end task of supervising or performing tumor localization simulations requires clinical judgment, multi-modal integration, and real-time decision-making that current AI systems cannot reliably handle independently at the required safety threshold.
Task automatability	claude-sonnet-5	1/5	This requires hands-on patient positioning, real-time imaging supervision, and clinical judgment in a physical clinical setting that current AI cannot perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Medical dosimetrists and radiologists are regulated professionals; clinical liability for dosimetry errors is high and falls on the supervising clinician or institution, creating strong legal and regulatory barriers to full automation regardless of technical capability.
Adoption barriers	claude-sonnet-5	5/5	Clinical imaging and tumor localization for radiation therapy planning require licensed, credentialed personnel and strict regulatory/safety oversight, making autonomous AI performance legally and clinically infeasible.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI image analysis tools require integration, validation, radiologist/dosimetrist oversight, and maintenance; the total cost of AI inference, integration, and mandatory human verification approaches or exceeds the cost of direct human performance of the task.
Cost vs. human wage	claude-sonnet-5	1/5	The physical simulation and patient interaction still require a trained human on-site, so AI cannot substitute for the labor cost of the core task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI tools for medical image segmentation and analysis exist in research and limited clinical settings, but no deployed product reliably performs complete tumor localization simulation supervision without substantial human oversight; error rates and narrow scope relative to the full clinical task limit production deployment.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously performs or supervises tumor localization simulations; AI segmentation tools exist but require a human dosimetrist to run and oversee the imaging session.

Advise oncology team members on use of beam modifying or immobilization devices in radiation treatment plans.

CI 3–25 · exposure 13 · augmentation 63 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Radiation oncology is conservative and safety-critical; adoption of AI for advisory clinical decisions remains slow outside research centers. Most institutions still rely on human dosimetrist expertise for device recommendations without AI augmentation in production.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare and radiation oncology adopt AI cautiously due to safety-critical nature and regulatory scrutiny, with clinical decision-support tools still in early, supervised deployment stages.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully surface device options, compare literature recommendations, and flag anatomical constraints, helping dosimetrists work faster and more comprehensively. However, the core advisory and judgment role remains human-centered, limiting transformative augmentation.
Augmentation potential	claude-sonnet-5	4/5	AI-based treatment planning systems and dose optimization algorithms already assist dosimetrists by suggesting beam arrangements and flagging immobilization needs, meaningfully boosting their planning efficiency.
Task automatability	claude-haiku-4-5-20251001	2/5	This task requires domain expertise, clinical judgment about patient anatomy and treatment goals, and integration with team workflows. While AI can suggest device parameters based on treatment plan data, the final advisory role—explaining trade-offs and coordinating with the oncology team—demands human expertise and accountability that current systems cannot reliably replace end-to-end.
Task automatability	claude-sonnet-5	1/5	This requires real-time clinical judgment integrating patient anatomy, treatment intent, and physics constraints in a collaborative team setting, which current AI cannot autonomously perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Radiation oncology is heavily regulated (FDA, state licensing), and dosimetrists must be credentialed professionals. Advising on treatment devices carries significant liability; regulatory frameworks and institutional quality-control protocols require human dosimetrist sign-off, creating strong legal and organizational barriers to full automation.
Adoption barriers	claude-sonnet-5	5/5	Radiation treatment planning decisions require credentialed clinical professionals and are subject to strict medical, regulatory, and liability oversight, making autonomous AI advice legally and clinically unacceptable.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Dosimetrists' advice is highly specialized; the cost of AI oversight, validation, and integration into clinical workflows likely exceeds the savings from partial automation. Liability and quality assurance costs remain high.
Cost vs. human wage	claude-sonnet-5	1/5	Since no reliable autonomous AI product performs this advisory function, there is no viable AI cost basis to compare against a dosimetrist's wage for this specific task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI can assist with literature-based device recommendations and parameter suggestions (demonstrated in some research prototypes), but no production system reliably advises clinical teams on complex, patient-specific device selection with the consistency required for patient safety in radiation oncology.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product independently advises oncology teams on beam modifiers or immobilization devices; existing planning software is a tool used by dosimetrists, not a substitute advisor.

Measure the amount of radioactivity in patients or equipment, using radiation monitoring devices.

CI 5–23 · exposure 13 · augmentation 38 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Healthcare, especially radiation oncology, has conservative adoption patterns for safety-critical tasks; dosimetry measurements remain tightly controlled by licensing requirements and institutional protocols, limiting automation incentives.
Sector adoption velocity	claude-sonnet-5	1/5	Radiation oncology and medical physics are highly regulated, equipment-dependent fields with slow AI adoption for physical measurement tasks specifically.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI could assist by automating data logging, flagging anomalous readings, and generating reports, but the dosimetrist must remain actively engaged in measurement execution and interpretation given the safety-critical nature of radiation dosimetry.
Augmentation potential	claude-sonnet-5	2/5	AI could assist with logging, trend analysis, or flagging anomalous readings from monitoring devices, but offers minimal help with the physical act of measurement itself.
Task automatability	claude-haiku-4-5-20251001	2/5	Radiation measurement itself (device reading) is straightforward, but interpreting readings in clinical context, positioning equipment correctly, and quality assurance require human judgment and expertise that current AI systems cannot reliably handle end-to-end.
Task automatability	claude-sonnet-5	1/5	This requires physically operating radiation monitoring devices (Geiger counters, ion chambers) on patients or equipment in a clinical setting, a hands-on physical measurement task AI cannot perform end-to-end today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Medical dosimetry is heavily regulated by the Nuclear Regulatory Commission and state agencies; a licensed professional must perform and sign off on radiation measurements for patient safety, creating a hard legal requirement for human involvement.
Adoption barriers	claude-sonnet-5	4/5	Radiation safety measurements are governed by regulatory and licensing requirements (NRC/state radiation safety programs) and typically require certified personnel, creating strong compliance barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Radiation monitoring devices are capital-intensive and require specialized maintenance; the cost of integrating AI monitoring plus human oversight would not undercut the operational cost of a trained dosimetrist performing these measurements.
Cost vs. human wage	claude-sonnet-5	1/5	AI cannot substitute for the physical measurement device and human operator, so there is no comparable AI-based cost structure for this task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While radiation detectors can be automated to log readings, no deployed AI system independently performs the full dosimetry measurement task including clinical decision-making and equipment validation in production dosimetry workflows.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product physically measures radioactivity levels; this remains a manual task performed by trained personnel using calibrated instruments.

Conduct radiation oncology-related research, such as improving computer treatment planning systems or developing new treatment devices.

CI 7–16 · exposure 5 · augmentation 75 · importance 3.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Research adoption in academic oncology centers is slow and cautious; most research remains fundamentally human-driven with AI used only for computation or data management. Clinical translation of new treatment planning systems requires regulatory approval and physician oversight, limiting pure automation.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare and clinical physics research adopts AI tools cautiously due to regulatory and safety concerns, though computational modeling and simulation tools are gradually being integrated.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI substantially assists dosimetrists in research by accelerating computational modeling, analyzing treatment planning datasets, optimizing algorithm parameters, and identifying patterns in clinical outcomes. These tools meaningfully amplify research productivity while the researcher retains control of hypothesis and direction.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist in literature review, data analysis, simulation of treatment plans, and computational modeling, accelerating parts of the research process even though humans drive the overall research direction.
Task automatability	claude-haiku-4-5-20251001	1/5	This task requires original research design, hypothesis formulation, experimental methodology, and domain expertise in radiation physics and oncology. Current AI systems lack the autonomous research capability, scientific creativity, and specialized knowledge integration needed to conduct novel research end-to-end.
Task automatability	claude-sonnet-5	1/5	This is open-ended scientific research and device/software innovation requiring novel experimentation, clinical validation, and engineering design—far beyond current AI's ability to execute end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Research integrity, institutional review board approval, and regulatory validation of new treatment systems require licensed professionals and human accountability. Publication and clinical adoption of findings demand human expertise and professional credibility that AI cannot substitute.
Adoption barriers	claude-sonnet-5	4/5	Research involving treatment devices and clinical planning systems is subject to significant regulatory, safety, and institutional review requirements, and typically requires credentialed medical physicists/dosimetrists to validate outputs.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-assisted research tools (simulation, analysis) cost significantly less per inference, but integrating them into productive research workflows requires substantial setup, oversight, and human expertise. The all-in cost remains comparable to or higher than human researcher time for novel research outcomes.
Cost vs. human wage	claude-sonnet-5	1/5	AI cannot substitute for the human researcher's role in designing experiments, validating clinical safety, and iterating on physical devices, so there is no meaningful cost-equivalent AI solution.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products autonomously conduct radiation oncology research. AI can assist with literature review, data analysis, or simulation validation, but research direction, experimental design, and clinical interpretation remain human-dependent. Only narrow research support tools exist in production.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously conducts radiation oncology research or develops new treatment devices; this remains a human-led, research-stage endeavor with AI only as a minor tool.

Fabricate beam modifying devices, such as compensators, shields, and wedge filters.

CI 5–10 · exposure 0 · augmentation 25 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Medical dosimetry is a specialized, low-digitization domain with small workforce numbers. Adoption of automation in this niche clinical manufacturing setting is extremely limited; the sector relies on traditional craftsmanship and human expertise.
Sector adoption velocity	claude-sonnet-5	1/5	Medical device fabrication in radiation oncology is a highly specialized, low-digitization physical process with minimal AI adoption reported in production settings.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI could marginally assist through design optimization, material simulation, or CAD automation, but the core task of physically fabricating these devices offers limited opportunity for AI-augmented human productivity. Most value remains with the skilled technician's hands-on work.
Augmentation potential	claude-sonnet-5	2/5	AI-assisted CAD/treatment planning software can help design specifications for compensators or shields, offering some productivity gains, but the fabrication itself remains manual.
Task automatability	claude-haiku-4-5-20251001	1/5	Fabricating physical beam-modifying devices requires precision manufacturing, material handling, and real-world assembly that current AI systems cannot perform end-to-end. While AI may assist in design or simulation, the actual fabrication involves specialized equipment and hands-on craftsmanship that remains outside the scope of deployed autonomous systems.
Task automatability	claude-sonnet-5	1/5	This is a physical fabrication task requiring hands-on machining or 3D-printing of custom devices tailored to patient anatomy, which current AI systems cannot perform end-to-end.rationale continues: AI may assist in designing device specifications but cannot physically fabricate them.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory barriers are substantial: fabricated devices must meet strict radiation safety standards and FDA/medical device regulations. Quality assurance, material traceability, and legal liability for device failures create strong protection against full automation, requiring human expertise and accountability.
Adoption barriers	claude-sonnet-5	3/5	While not licensed to a specific individual by law in the same way as prescribing, quality/safety requirements in radiation therapy impose strong oversight and validation processes before devices are used clinically.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The cost of acquiring, maintaining, and programming automated fabrication equipment far exceeds the cost of employing a skilled dosimetrist for this specialized manufacturing task. Current AI cannot reduce the capital and operational costs below those of direct human labor.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI-only substitute for physical fabrication, so cost comparison favors the human-operated equipment and technician labor already required, with no AI displacing that cost.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed product exists that can autonomously fabricate physical radiation therapy devices like compensators and shields. This task requires specialized industrial equipment, material expertise, and quality assurance that is performed by trained human technicians, not by AI systems.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product fabricates physical beam-modifying devices; this remains a manual/technician-operated manufacturing process using milling machines or 3D printers under human control.

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.