Physical Therapist Assistants
31-2021.00Median wage $68,380/yr112,430 employed (US)Rank #788 of 923 scored · top 85% by substitution
Assist physical therapists in providing physical therapy treatments and procedures. May, in accordance with state laws, assist in the development of treatment plans, carry out routine functions, document the progress of treatment, and modify specific treatments in accordance with patient status and within the scope of treatment plans established by a physical therapist. Generally requires formal training.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution15
Exposure13
Augmentation39
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
21 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
0%
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%14
panel mean rating 1.6/5 → substitution pressure 14/100
Technical feasibility todayw 20%12
panel mean rating 1.5/5 → substitution pressure 12/100
Cost vs. human wagew 15%13
panel mean rating 1.5/5 → substitution pressure 13/100
Adoption barriersw 20%inverted — strong barriers lower the score24
panel mean rating 4.0/5 (barrier strength) → substitution pressure 24/100
Sector adoption velocityw 10%13
panel mean rating 1.5/5 → substitution pressure 13/100
Task breakdown (21 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.
Perform clerical duties, such as taking inventory, ordering supplies, answering telephone, taking messages, or filling out forms.
66CI 56–76 · exposure 62 · augmentation 63 · importance 3.1/5 · click for rater detail
Perform clerical duties, such as taking inventory, ordering supplies, answering telephone, taking messages, or filling out forms.
66| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Healthcare organizations show moderate adoption of clerical automation (scheduling bots, intake forms), but rollout remains uneven due to legacy system integration challenges and staffing inertia; adoption is faster in larger systems than small practices. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare administrative back-office functions are adopting automation, but clinical/allied health settings, especially smaller PT practices, lag behind faster-digitizing sectors like finance or general office work. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI assistants can usefully augment clerical work by auto-suggesting inventory items, pre-filling forms, and flagging urgent messages, improving assistant productivity while the human remains responsible for final decisions and quality control. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can meaningfully speed up clerical tasks such as drafting forms, managing inventory logs, and triaging calls/messages, freeing the assistant to focus on patient care while remaining in the loop. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Most clerical duties (inventory tracking, supply ordering, form filling, message taking) are highly automatable with current AI and RPA systems. Document processing and basic telephone answering are mature technologies, though some contextual judgment in supply ordering may require human oversight, preventing a full rating of 5. |
| Task automatability | claude-sonnet-5 | 3/5 | Sub-tasks like answering phones, taking messages, and inventory tracking can be substantially automated with current AI (chatbots, IVR, inventory software), but filling out clinical forms and coordinating physical supplies still require human involvement, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or legal barriers exist for automating clerical work in physical therapy settings. Integration with existing EHR systems and organizational workflows presents some friction, but no licensing requirement or liability asymmetry prevents substitution. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensure is required for clerical work itself, though some forms may require signature by a licensed PTA, and clinics may prefer human staff for patient-facing communication, creating mild friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-based clerical automation is an order of magnitude cheaper than human labor per task equivalent; RPA and chatbot costs are minimal compared to the loaded wage of an assistant performing these repetitive duties. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated phone systems, inventory software, and form-filling tools are inexpensive relative to a clinician's or assistant's loaded wage for repetitive clerical work. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products reliably perform these tasks in healthcare and other sectors: RPA platforms handle forms and inventory systems, chatbots manage routine phone inquiries, and document AI processes supplies data. Minor limitations exist in complex multi-step ordering decisions, but production systems are widely used. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Deployed products (scheduling/phone assistants, inventory management systems, EHR-integrated forms) exist and are used in healthcare settings, but they are not universally reliable across all clerical sub-tasks in small clinical practices. |
Document patient information, such as notes on their progress.
43CI 25–60 · exposure 45 · augmentation 88 · importance 4.9/5 · click for rater detail
Document patient information, such as notes on their progress.
43| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare adoption of autonomous documentation AI remains cautious. While voice-to-text and template tools are used, they are assistive rather than replacement systems. Regulatory risk, liability concerns, and the high cost of errors in medical records slow deployment of truly autonomous solutions. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare, especially outpatient rehab and PT clinics, is a slower-adopting sector for AI tools relative to information/finance industries, with pilots more common than full production use. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-assisted tools—voice transcription, prompt-based templates, auto-population of structured fields—significantly boost PTA documentation efficiency by reducing typing and organizing information. These systems keep the human in the loop for clinical validation while meaningfully raising throughput. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI scribes and note-drafting assistants meaningfully speed up documentation while the PTA remains responsible for reviewing and finalizing clinical content. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate draft summaries from structured clinical data or voice notes, documenting patient progress requires nuanced clinical judgment about what is clinically significant, accurate capture of subjective patient reports, and adherence to medical-legal standards. Current AI alone cannot reliably determine what should be documented without substantial human review and correction. |
| Task automatability | claude-sonnet-5 | 4/5 | AI can draft progress notes from structured inputs or dictation with substantial time savings, though a human must verify clinical accuracy before finalizing.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Medical records must be accurate and defensible in legal/regulatory contexts; PTA notes are part of the legal medical record. Many jurisdictions and institutional policies require a licensed clinician to review and sign clinical documentation, creating a hard supervisory requirement that prevents full substitution. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Documentation must meet payer and regulatory (Medicare/insurance) standards and often requires clinician sign-off, creating moderate compliance friction even though the task itself isn't licensed exclusively for humans. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI documentation tools (speech recognition, template filling) can reduce typing time but require oversight, verification, and correction by licensed staff. Total cost including integration, quality control, and liability mitigation remains comparable to direct human documentation, especially given error-correction overhead. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-assisted documentation tools are inexpensive per note compared to the loaded time cost of a PTA manually writing detailed progress notes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Voice-to-text and templated documentation systems exist in production, but they require significant human editing to ensure accuracy, clinical appropriateness, and legal compliance. No system reliably produces deployment-ready clinical notes autonomously; clinicians cannot fully trust unsupervised AI output for medical records. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Ambient documentation and clinical scribe tools are deployed in healthcare settings, but PT-specific documentation products are less mature than general medical scribes and require review for coding/compliance accuracy. |
Clean work area and check and store equipment after treatment.
36CI 19–52 · exposure 36 · augmentation 25 · importance 4.3/5 · click for rater detail
Clean work area and check and store equipment after treatment.
36| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | PT clinics and rehabilitation facilities are slower adopters of automation; most are small to mid-sized independent practices with limited capital budgets and conservative IT spending. Robotics adoption in clinical support functions remains rare outside large hospital systems. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical therapy clinics are low-digitization, hands-on environments with minimal robotic automation of housekeeping tasks; adoption of AI/robotics for this specific task is essentially nonexistent. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | Robotic vacuums and autonomous disinfection units can assist by handling routine cleaning, freeing the assistant for more focused equipment checks and organization. However, the augmentation is partial—human judgment on damage detection and proper storage remains essential. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI tools offer little to no assistance for the physical acts of cleaning and storing equipment, as this is a manual task with no meaningful digital or cognitive component to augment. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | A mobile robot with grasping capability and environmental scanning could handle a large share of equipment storage, sanitization, and area cleaning. However, inspection for damage and safe storage placement for specialized equipment may require human judgment, preventing full end-to-end automation, though 50%+ time savings is achievable. |
| Task automatability | claude-sonnet-5 | 2/5 | This is a physical manipulation and housekeeping task requiring mobility, dexterity, and object handling in a clinical space; current general-purpose AI cannot perform it end-to-end without specialized robotics. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Healthcare facilities face regulatory requirements around disinfection standards and equipment sterilization; automated systems must meet compliance, and liability concerns exist if robotic equipment storage damages specialized tools. Customer comfort and clinic tradition favor human oversight, creating moderate friction to adoption. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No strict licensing requirement governs this specific cleaning/storage subtask, though infection control and equipment safety protocols in clinical settings create some procedural expectations for human accountability. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current robotic systems capable of cleaning and equipment handling carry high capital and maintenance costs ($50k–$150k+) that are difficult to justify for a single part-time task in small-to-medium PT clinics. Loaded labor cost for assistants ($20–$30/hour) often remains competitive on a per-task basis. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | No viable AI/robotic system exists at scale for this task, so any hypothetical automation would require expensive specialized robotics far costlier than a human aide performing the same work. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While robotic cleaning exists and some autonomous sanitization systems are deployed in hospitals, integrated end-to-end cleaning and equipment-checking workflows tailored to PT clinics are not yet reliably performed at scale by commercial products. Most deployed systems handle single functions (vacuuming or disinfection), not the full task. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs clinical equipment cleaning, checking, and storage in physical therapy settings; this remains outside current robotic deployment in healthcare facilities. |
Monitor operation of equipment and record use of equipment and administration of treatment.
28CI 25–30 · exposure 25 · augmentation 50 · importance 4.0/5 · click for rater detail
Monitor operation of equipment and record use of equipment and administration of treatment.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare adoption of AI monitoring remains slow and largely in pilot phases; physical therapy clinics are typically small-to-medium organizations with limited digital infrastructure and conservative adoption patterns around patient-facing clinical tasks. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare, especially allied health/physical therapy settings, has historically been slower to adopt AI compared to information/finance sectors, with most current use limited to administrative documentation pilots. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-assisted documentation tools and sensor integration can usefully help assistants log treatment details and flag equipment issues, reducing manual charting burden while the human maintains oversight of patient safety and clinical appropriateness. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-powered documentation and voice transcription tools can meaningfully speed up record-keeping of treatment administration, letting the assistant focus more on hands-on equipment monitoring and patient care. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI could potentially log equipment usage and basic treatment records through computer vision or sensor integration, the task requires real-time monitoring for safety and clinical appropriateness, which demands human judgment about patient condition and equipment malfunction that current AI cannot reliably provide end-to-end. |
| Task automatability | claude-sonnet-5 | 2/5 | Recording equipment use and treatment administration could be partially automated via structured data entry or voice-to-text, but real-time monitoring of patient response to equipment requires physical presence and clinical judgment.24 The documentation portion alone doesn't meet the full 50% threshold for the combined task. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Healthcare regulations require licensed or certified personnel to oversee patient treatment; liability concerns around equipment safety and patient monitoring create a strong requirement for human accountability, and clinical settings have high error-cost sensitivity. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement mandates a human for equipment monitoring, but patient safety concerns, liability for treatment errors, and the hands-on nature of therapy create moderate friction against automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI systems for clinical monitoring and documentation still require significant infrastructure, integration, and human oversight, making the all-in cost comparable to or higher than a therapist assistant's wage for this specific task. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI documentation tools have some cost advantage for the recording piece, but the equipment monitoring requires an in-person assistant, so overall cost savings versus the human PTA are limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Some sensor-based logging and basic video analysis products exist for equipment tracking, but no deployed system reliably monitors equipment operation for clinical safety and simultaneously records treatment administration with the accuracy and contextual judgment required in physical therapy settings. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | EHR systems and some clinical documentation assistants exist, but no deployed product autonomously monitors physical therapy equipment operation in real clinical settings; documentation aids are narrow and require human oversight. |
Communicate with or instruct caregivers or family members on patient therapeutic activities or treatment plans.
23CI 16–30 · exposure 17 · augmentation 63 · importance 4.3/5 · click for rater detail
Communicate with or instruct caregivers or family members on patient therapeutic activities or treatment plans.
23| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare sectors show slow, cautious adoption of AI for patient-facing communication due to liability concerns, regulatory oversight, and organizational preference for licensed staff in therapeutic relationships. Pilots exist but production displacement is minimal; most healthcare systems maintain human practitioners in direct caregiver instruction roles. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare, especially hands-on rehabilitative care, has been slower to adopt AI-driven patient communication tools compared to purely digital sectors like finance or software. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by drafting patient education materials, generating exercise illustrations, or summarizing treatment plans for the PTA to present—useful productivity gains in preparation. However, the core act of live instruction, answering caregiver questions, and addressing concerns still requires the human PTA, so augmentation is partial rather than transformative. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully help PTAs by generating personalized handouts, exercise instructions, and translated materials, improving efficiency and clarity while the assistant remains the primary communicator. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires personalized communication adapted to individual family contexts, emotional support, and real-time clarification—core human interpersonal competencies that current AI systems cannot reliably perform end-to-end. While AI can draft instructional materials, the actual communication and instruction with real caregivers demands presence, empathy, and responsiveness to feedback that AI lacks today. |
| Task automatability | claude-sonnet-5 | 2/5 | While AI could draft educational materials or summaries, the actual real-time communication, answering questions, and building trust with caregivers requires human presence and adaptive interpersonal judgment that current systems cannot fully replace. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Scope-of-practice regulations require licensed PTAs to provide patient education and caregiver instruction; liability and clinical accountability rest on the qualified human practitioner. Many healthcare organizations have explicit policies mandating direct practitioner-to-family communication for treatment plans, creating organizational and legal barriers to full substitution. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement mandates this specific communication be done by a PTA, but liability, patient trust, and the need for tailored clinical judgment create meaningful friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI costs for document generation or chatbots are low in isolation, but integration, customization per patient, and required human oversight to ensure safety and appropriateness make the total system cost comparable to or higher than direct PTA communication time. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply generate written instructions, but the live conversational and relational component still requires a paid clinician, keeping overall cost comparable to human-only delivery. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed products reliably perform live instruction or therapeutic counseling of caregivers at clinical scale. AI can generate template instructions or informational content, but these require human PTA review and delivery; no production system independently conducts this caregiver-facing communication with clinical accountability. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some patient education chatbots and AI-generated instructional handouts exist, but no deployed product reliably conducts the full caregiver communication and instruction process in clinical practice. |
Attend or conduct continuing education courses, seminars, or in-service activities.
19CI 13–25 · exposure 17 · augmentation 50 · importance 3.9/5 · click for rater detail
Attend or conduct continuing education courses, seminars, or in-service activities.
19| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare professions have begun adopting online and hybrid learning formats, but regulatory requirements for documented human attendance at CE activities, combined with professional standards emphasizing live instruction, limit rapid displacement of the attendance requirement itself. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare licensing and CE compliance sectors are slow to adopt AI-driven changes to mandatory training structures, though online CE format is already common. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by summarizing course materials, generating study guides, or helping organize and review continuing education content, providing useful support to the learner, though the core task of attending and engaging remains human-centered. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help identify relevant courses, summarize material, generate practice quizzes, or assist those conducting training with content creation, improving efficiency around the task. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires human presence, engagement, and interaction with learning material and instructors that fundamentally cannot be automated end-to-end. AI cannot attend courses on behalf of a person or conduct seminars requiring real-time group facilitation and participant responsiveness. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can help find, summarize, or curate continuing education content, but the core act of attending or conducting a course/seminar requires physical or live human presence and participation.19 It cannot be fully substituted end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory and licensing bodies typically mandate that licensed professionals like physical therapist assistants complete continuing education hours through accredited courses and in-person seminars, with documentation of attendance and participation required for license maintenance. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Continuing education is typically mandated by state licensing boards with specific credit-hour and attendance/participation requirements, creating a hard regulatory barrier to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of AI-generated or AI-delivered training would not eliminate the need for a human physical therapist assistant to attend and engage with continuing education, so there is no meaningful cost displacement; the human expense remains mandatory. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply produce or summarize training materials, but actual course attendance/certification still requires human time investment, so overall cost savings are limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can generate educational content or deliver some automated training modules, it cannot meaningfully substitute for attending live seminars or conducting interactive in-service activities where human presence and participation are essential elements of the task. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Products exist for online CE course delivery and AI-assisted content summarization, but no deployed system 'attends' or 'conducts' the training on behalf of the person in a way that satisfies certification requirements. |
Instruct patients in proper body mechanics and in ways to improve functional mobility, such as aquatic exercise.
16CI 7–25 · exposure 13 · augmentation 50 · importance 4.7/5 · click for rater detail
Instruct patients in proper body mechanics and in ways to improve functional mobility, such as aquatic exercise.
16| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare adoption of AI in direct patient care remains slow and cautious due to regulatory, liability, and quality assurance requirements; physical therapy is a human-contact service sector with strong organizational preference for licensed practitioners. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare/rehab settings are historically slow AI adopters for hands-on physical care tasks, with adoption concentrated in administrative or documentation functions rather than direct patient instruction. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by generating exercise videos, tracking patient progress through motion capture, or drafting exercise program outlines, but the core task of live instruction and form correction remains human-dependent for safety and clinical judgment. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can supply exercise libraries, generate patient education materials, and offer video-based movement analysis to support the PTA's instruction, improving efficiency without replacing the interaction. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires real-time observation of patient movement, immediate corrective feedback, hands-on physical guidance, and adaptation to individual biomechanics—core competencies of in-person care that current AI cannot perform reliably or safely without a human present. |
| Task automatability | claude-sonnet-5 | 2/5 | Instruction requires live physical observation, hands-on correction, and adaptive coaching of body mechanics that current AI cannot perform end-to-end without a human physically present. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | State licensure laws require licensed physical therapists to supervise patient care and directly oversee therapeutic instruction; PTA scope of practice is legally defined and involves hands-on patient contact that cannot be delegated to non-licensed automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Clinical supervision requirements, liability for patient injury during exercise instruction, and licensure/scope-of-practice rules create strong barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The loaded cost of a physical therapist assistant (wages + benefits ~$35–50k/year) remains lower than the integrated cost of AI-generated personalized instruction with required human supervision, liability management, and quality oversight. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI-generated educational content is cheap, but the actual hands-on instruction and safety supervision still require a human PTA, keeping overall costs comparable to or higher than pure automation claims. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can generate instructional videos and written guidance on body mechanics and aquatic exercises, no deployed product reliably evaluates patient form, provides real-time corrective feedback, or adapts instruction to patient limitations in a clinical setting without human oversight. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product provides autonomous in-person physical instruction and correction of patient movement; this remains research-stage or purely conceptual (e.g., video-based feedback apps with narrow scope). |
Confer with physical therapy staff or others to discuss and evaluate patient information for planning, modifying, or coordinating treatment.
16CI 7–25 · exposure 13 · augmentation 50 · importance 4.5/5 · click for rater detail
Confer with physical therapy staff or others to discuss and evaluate patient information for planning, modifying, or coordinating treatment.
16| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Physical therapy remains a relationship-intensive, regulated healthcare sector with slow AI adoption in clinical workflows. While documentation and scheduling tools exist, production deployment of AI in care coordination conferences is minimal—most clinics continue to rely on manual team discussions and human note-taking. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare, especially allied health and physical therapy settings, has historically lagged in AI adoption for clinical coordination tasks compared to information/finance sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist by automatically compiling patient history, flagging relevant prior assessments, and suggesting evidence-based modifications, allowing therapists to focus discussion on clinical judgment. However, the augmentation is primarily informational rather than transformative, as the core collaborative evaluation remains human-driven. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist by summarizing patient records, flagging trends in progress data, or drafting treatment notes to inform the discussion, meaningfully aiding preparation even though the conferral itself remains human-led. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can summarize patient information and suggest treatment modifications based on clinical guidelines, the collaborative discussion and real-time clinical judgment required to evaluate patient progress and coordinate multidisciplinary care cannot be reliably automated end-to-end today. AI currently plays a supporting role in information synthesis rather than replacing the interactive clinical conference itself. |
| Task automatability | claude-sonnet-5 | 1/5 | This is an interpersonal, judgment-based clinical collaboration task requiring physical presence, hands-on assessment context, and professional discretion that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: clinical decision-making and care coordination typically require a licensed physical therapist's judgment and accountability, patient privacy regulations (HIPAA) restrict data sharing, and professional liability concerns make autonomous AI coordination of patient treatment risky without human sign-off. Legal and regulatory frameworks expect human professionals to own treatment decisions. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Clinical treatment planning involves licensed practitioner judgment, liability for patient outcomes, and regulatory/professional standards requiring qualified human staff to confer and sign off on care decisions. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI systems for clinical summarization and decision support require significant integration, human oversight to validate recommendations, and ongoing model maintenance, making the all-in cost comparable to or exceeding the time saved from a junior PT assistant's labor on this collaborative task. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute performing this interpersonal coordination task, so cost comparison favors the human by default. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs the full coordination and evaluation conference autonomously. AI tools exist for documentation summarization and clinical decision support, but they cannot replace the interpersonal judgment, nuanced patient assessment, and dynamic team communication that this task demands in clinical settings. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product conducts clinical team conferrals or coordinates treatment decisions autonomously; at best AI tools support documentation review, not the conferral itself. |
Prepare treatment areas and electrotherapy equipment for use by physiotherapists.
16CI 5–26 · exposure 8 · augmentation 13 · importance 3.9/5 · click for rater detail
Prepare treatment areas and electrotherapy equipment for use by physiotherapists.
16| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare, particularly outpatient physical therapy, remains low-digitization with heavy reliance on human presence. Very limited robotics adoption in PT clinics; industry shows no meaningful trend toward automation of equipment preparation tasks. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare physical settings adopt automation slowly, and equipment prep tasks in physical therapy have seen negligible AI or robotic adoption to date. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist with checklists, maintenance reminders, or equipment availability tracking, but provides limited augmentation for the core physical setup work. Any assistance would be informational rather than productivity-transforming for the hands-on aspects of the task. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI offers minimal assistance for physically setting up rooms and machines; there's no meaningful software layer that speeds up this manual task. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI could theoretically assist with organizing checklists or scheduling equipment maintenance, the physical setup of treatment areas and calibration of electrotherapy devices requires hands-on manipulation, positioning, and safety verification that current AI systems cannot perform in situ. Some planning components could be automated, but the majority requires physical presence and real-time adjustment. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical setup task requiring manual handling of equipment, positioning, and sanitation in a clinical space—current AI systems cannot perform physical manipulation of treatment areas or electrotherapy devices. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Clinical settings require physical presence and hands-on safety checks before equipment use; liability for equipment malfunction or improper setup is high. Regulatory oversight of medical device handling and workplace safety standards create meaningful barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing strictly requires a human for equipment setup, but practical/physical constraints and clinical safety protocols create moderate friction against automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The capital and integration costs of robotics capable of handling electrotherapy equipment setup and room preparation far exceed the loaded wage of a PT assistant, making automation economically infeasible at current technology costs. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute for this physical task, so the human remains the only cost-effective option; any hypothetical robotic solution would be far more expensive than assistant labor. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No current deployed AI system can autonomously prepare physical treatment spaces and equipment in a clinical setting. This task fundamentally requires robotic manipulation with medical-grade reliability, which does not exist in production PT environments today. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs physical preparation of treatment rooms or electrotherapy equipment; this remains squarely a human physical task with no robotic solution in production. |
Observe patients during treatments to compile and evaluate data on their responses and progress and provide results to physical therapist in person or through progress notes.
15CI 5–25 · exposure 13 · augmentation 38 · importance 4.8/5 · click for rater detail
Observe patients during treatments to compile and evaluate data on their responses and progress and provide results to physical therapist in person or through progress notes.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare remains a laggard sector for autonomous AI automation due to high regulatory burden, low digitization of clinical observation, and strong institutional preference for human clinical judgment in patient monitoring. Adoption of AI in this specific task is negligible in production settings. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare, especially hands-on rehabilitative therapy settings, is a slower-adopting sector for AI relative to information/professional services, with pilots more common than production deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI can assist modestly with post-session note generation, data organization, or flagging outliers in patient metrics, but the core task of observing and evaluating patient responses requires human clinical expertise and presence, limiting augmentation value. AI tools are secondary to the main clinical assessment work. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-assisted transcription and note-generation tools can help PTAs draft progress notes faster after observation, offering moderate productivity benefits for the documentation portion of the task. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | The task requires real-time, in-person observation of patient physical responses, subtle behavioral cues, and clinical judgment about treatment efficacy—elements that current AI systems cannot reliably perform without human presence and assessment. Observation and evaluation are fundamentally synchronous, embodied activities requiring contextual understanding and adaptive decision-making that exceed today's capabilities. |
| Task automatability | claude-sonnet-5 | 2/5 | Physical observation of patients during hands-on treatment and real-time clinical judgment about their responses cannot be performed by AI; only the documentation portion is automatable, and that requires accurate human-generated input first. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | State licensing requirements and liability concerns create strong barriers: physical therapy licensure boards typically mandate that treatment decisions and patient monitoring be performed or directly supervised by licensed professionals, and clinical error carries significant patient-safety liability. The human-contact component and regulatory oversight of clinical assessment are substantial. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Clinical documentation of patient progress typically requires licensed/credentialed personnel for accuracy and liability, and hands-on treatment observation inherently requires human presence, creating strong regulatory and safety barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | An AI system capable of in-person observation and clinical evaluation would require specialized hardware, integration, and oversight costs that far exceed the salary of a physical therapist assistant. The task is also too specialized and lower-volume per site to achieve cost advantage. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI could cheaply assist with note drafting, but the core task requires a present human to observe and interact physically with the patient, so most of the labor cost remains unavoidable. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product can independently observe patients during physical therapy, evaluate their responses to treatment, and compile clinical assessments at production scale. While AI can assist with note documentation or data entry after human observation, it cannot replace the observational and evaluative components of this task. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Ambient documentation and note-generation tools exist in healthcare but are not widely deployed specifically for PTA observation-to-note workflows, and no product performs the physical observation component. |
Train patients in the use of orthopedic braces, prostheses, or supportive devices.
15CI 5–25 · exposure 13 · augmentation 38 · importance 4.2/5 · click for rater detail
Train patients in the use of orthopedic braces, prostheses, or supportive devices.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare adoption of AI automation remains cautious and heavily regulated. Physical therapy is a small, offline-intensive sector with strong professional norms favoring in-person care; automation pilots are rare and production deployment minimal in this specific clinical domain. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical therapy is a hands-on healthcare sector with low digitization for direct patient training tasks, showing minimal AI adoption for this specific activity. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI could assist by generating visual training materials, providing video walkthroughs, or offering standardized instructional templates that a PTA uses when training patients. This reduces preparation time but does not transform the core in-person training and fitting work. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could provide instructional videos, reminders, or exercise tracking apps to support patient education, but it offers limited assistance for the hands-on training itself. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Training on orthopedic devices requires hands-on fitting, real-time adjustment, and physical demonstration that AI cannot perform autonomously. While AI could provide instructional content or video guidance, the core task—ensuring proper fit, monitoring patient comfort, and adapting instruction to individual physical limitations—demands human presence and tactile feedback. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires hands-on physical demonstration, tactile correction of patient movement, and real-time adjustment based on the patient's body and comfort, which current AI cannot perform end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | State licensure laws typically require a licensed physical therapist or certified PTA to directly supervise or perform device fitting and training to ensure patient safety and liability protection. Insurance and healthcare regulations reinforce that a credentialed human must sign off on device fit and training. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Requires licensed clinical judgment, physical contact, and safety oversight for device fitting and gait/use training, creating strong professional and liability barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-generated instructional content is cheap, but it cannot replace the labor-intensive one-on-one fitting and adjustment sessions. The human labor cost dominates; AI supplementation saves modest time only on content creation, not on the critical hands-on work. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute delivering this physical training, so AI cost is not comparable; the human PTA remains the only viable cost option. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably trains patients in device use end-to-end. AI can generate instructional videos or educational materials, but these do not replace the supervised, individualized fitting and adjustment sessions that define the task in clinical practice. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product physically trains patients in device use; this remains a hands-on clinical task performed only by human clinicians. |
Measure patients' range-of-joint motion, body parts, or vital signs to determine effects of treatments or for patient evaluations.
14CI 5–23 · exposure 13 · augmentation 38 · importance 4.2/5 · click for rater detail
Measure patients' range-of-joint motion, body parts, or vital signs to determine effects of treatments or for patient evaluations.
14| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare settings, especially outpatient physical therapy clinics, have low adoption of autonomous measurement automation. Adoption remains pilot-stage; the sector is digitizing slowly and remains risk-averse regarding measurement automation due to regulatory and liability constraints. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical therapy is a hands-on, in-person healthcare service with low digitization of the core physical assessment process, and adoption of AI for direct physical measurement is essentially nonexistent. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-assisted measurement (e.g., computer vision for posture feedback, automated vital sign logging) can reduce documentation burden and provide real-time visual feedback during assessment, but the human PTA must remain in control of the hands-on examination and clinical decision-making. Modest productivity gains are plausible without replacing the human role. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with documentation, tracking trends in recorded measurements over time, or suggesting interpretations, but it does not meaningfully assist the physical act of measuring joint motion or vital signs itself. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While some vital sign measurement (e.g., heart rate, blood pressure) can be automated with wearable devices, most range-of-motion and joint measurement requires hands-on assessment, positioning judgment, and interpretation of subtle biomechanical feedback that current AI cannot replicate end-to-end. The task involves physical contact and real-time decision-making about patient positioning and compensation patterns. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires hands-on physical manipulation of patient joints and body parts using goniometers and manual palpation, which current AI systems cannot physically perform without embodied robotics that don't exist in deployable form for this purpose. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Physical therapists and their assistants operate under state licensure requirements and scope-of-practice regulations; in most jurisdictions, a licensed or credentialed human must perform or directly oversee patient assessment and measurement. Clinical liability, documentation requirements, and insurance billing standards create high organizational and regulatory barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Clinical measurement for treatment evaluation typically requires a licensed or certified professional performing hands-on assessment, with liability concerns around patient safety and accurate documentation feeding into care decisions. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Specialized sensor systems, calibration, and clinical oversight would be required to match human assistant performance. The equipment and integration costs would likely equal or exceed the hourly loaded wage of a PTA, particularly given the low error tolerance in healthcare measurement. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute performing this physical measurement task, so cost comparison favors the human by default since AI cannot deliver the output at all. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Some narrow components exist—standalone vital sign monitors and experimental computer vision systems for posture analysis—but no integrated product reliably performs full range-of-motion measurement or body-part assessment with clinical accuracy comparable to human physical therapist assistants. Deployed systems lack the tactile sensing and adaptive interaction needed for clinical reliability. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs hands-on physical measurement of range-of-motion or vital signs on patients in clinical PT settings; this remains a physical, in-person clinical task. |
Fit patients for orthopedic braces, prostheses, or supportive devices, such as crutches.
8CI 0–16 · exposure 8 · augmentation 38 · importance 3.8/5 · click for rater detail
Fit patients for orthopedic braces, prostheses, or supportive devices, such as crutches.
8| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare is still early in AI adoption overall, and hands-on clinical tasks like device fitting have slow adoption rates due to liability concerns, licensure, and the physical nature of the work. This task is in a laggard sector relative to information-intensive professional services. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare and physical rehabilitation settings, especially hands-on clinical tasks, show slower AI adoption compared to fully digital, information-based sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI could assist by analyzing gait video to suggest device type, predicting fit issues from patient measurements, or providing fitting protocols. However, assistance is limited to decision support since the actual fitting remains entirely manual and clinician-led. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could assist with measurement calculations, device selection recommendations, or fitting protocol documentation, but offers limited direct assistance during the physical fitting process itself. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Fitting orthopedic devices requires in-person measurement, physical assessment, and manual adjustment based on individual anatomy and gait patterns. AI cannot currently measure patients, assess fit quality through touch, or make real-time physical adjustments needed for proper device function. |
| Task automatability | claude-sonnet-5 | 1/5 | Fitting braces, prosthetics, or crutches requires hands-on physical manipulation, body measurement, and real-time tactile adjustment that current AI systems cannot perform without robotic embodiment, which is not deployed for this task. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Physical therapy assistants must work under licensed physical therapist supervision, and device fitting carries liability risk if poorly executed (risk of injury, poor function, skin breakdown). Regulatory requirements and the need for in-person physical manipulation create hard barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | This requires physical patient contact, clinical judgment about fit and safety, and often falls under licensed practice scope, creating strong practical and regulatory barriers to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Current AI systems cannot reduce the cost of fitting since the task fundamentally requires a trained human to be physically present with the patient. Deployed systems cannot perform this task, so there is no cost displacement. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing the physical fitting task, so AI cost is not comparable—human labor is currently the only option. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product performs end-to-end device fitting today. While AI can support measurement via imaging or suggest device types, the actual fitting process—which is tactile, iterative, and requires real-time patient feedback—remains entirely manual and requires human presence. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously fits orthopedic devices on patients today; this remains a physical, hands-on clinical task performed by trained personnel. |
Instruct, motivate, safeguard, and assist patients as they practice exercises or functional activities.
4CI 0–7 · exposure 0 · augmentation 38 · importance 4.9/5 · click for rater detail
Instruct, motivate, safeguard, and assist patients as they practice exercises or functional activities.
4| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare, especially physical rehabilitation, is a heavily regulated, human-contact sector with low automation adoption for direct patient care. Current adoption of AI in this space is limited to administrative or diagnostic support, not patient-facing assistance. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare, especially hands-on rehabilitation therapy, has been slow to adopt AI for direct physical patient care compared to administrative or diagnostic support functions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist by providing exercise video demonstrations or real-time form feedback via computer vision, but the core task—physical safeguarding, hands-on adjustment, and in-person motivation—remains fundamentally human. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist via motion-tracking apps, exercise reminder systems, or generating personalized exercise plans and motivational content, but cannot replace the physical safeguarding component. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task fundamentally requires real-time physical presence, tactile feedback, safety monitoring, and adaptive motivation based on patient response—all beyond current AI capabilities. No AI system can physically guard a patient or adjust body mechanics through touch. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires hands-on physical spotting, manual assistance, and real-time physical safeguarding of patients performing exercises, which current AI systems cannot perform as they lack physical embodiment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | This task has hard regulatory and legal barriers: physical therapists are licensed professionals who must supervise patient care, patients require hands-on assistance and contact, and liability for patient injury during exercise is substantial and non-delegable to machines. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Patient safety during physical activity requires a trained human present to prevent falls or injury, and liability concerns plus licensing/supervision requirements under PT oversight create strong barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of deploying a physical system with the dexterity, responsiveness, and safety certification needed to replace a human PTA would vastly exceed the loaded wage of the human assistant. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot substitute for the physical presence and manual assistance required, so no meaningful cost comparison favors AI for the core physical task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product reliably performs this task end-to-end. While AI can provide exercise instructions via video or text, it cannot physically assist, manually correct form, catch a falling patient, or provide the real-time safety supervision this role demands. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product physically safeguards or assists patients during exercise; this remains a physical-world task requiring human presence and touch. |
Secure patients into or onto therapy equipment.
3CI 0–5 · exposure 0 · augmentation 13 · importance 4.6/5 · click for rater detail
Secure patients into or onto therapy equipment.
3| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Physical therapy remains a hands-on, low-digitization field with strong preference for human contact and trust. Adoption of automation in patient handling is minimal; most clinics use manual techniques and equipment designed for human operation. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Healthcare physical therapy settings show minimal adoption of robotics for direct patient handling tasks; the sector lags in physical automation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could potentially assist with positioning guidance via computer vision (e.g., recommending optimal angles), but the core task—physically securing patients—requires human judgment and direct handling. Augmentation potential is limited to advisory rather than transformative. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI offers essentially no meaningful assistance for the physical act of positioning and securing a patient on equipment. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical manipulation of patients' bodies and positioning them safely on equipment—work that demands tactile feedback, real-time adjustment to individual patient needs, and direct human contact. No current AI system can physically handle patients or adapt to their responses in real time. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a hands-on physical task requiring manual manipulation of patients and equipment, which current AI systems cannot perform; no robotic system can safely and reliably secure a patient onto therapy equipment today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Patient safety, liability, and medical negligence law create hard barriers. A licensed human (PT or PTA) must legally ensure proper patient securing to avoid injury; automation would face regulatory scrutiny and cannot replace the human sign-off on patient safety. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Patient safety, physical contact, and clinical oversight requirements create strong practical barriers, though not always a strict licensure mandate specifically for this sub-task. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Robotic systems capable of safely handling patients are extremely expensive, require specialized infrastructure, and need continuous oversight. The loaded cost of a physical therapist assistant is far lower than the capital and operational cost of such a system. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI/robotic substitute for this physical task, so any hypothetical automation would require expensive specialized robotics far costlier than a human assistant. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product performs patient securing or equipment positioning autonomously today. The task requires embodied robotics with force control, safety assurance, and medical liability considerations that remain research-stage only. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs physical patient handling and securing tasks in clinical PT settings; this remains firmly in the physical/robotics domain, far from current AI capabilities. |
Assist patients to dress, undress, or put on and remove supportive devices, such as braces, splints, or slings.
3CI 0–5 · exposure 0 · augmentation 13 · importance 3.9/5 · click for rater detail
Assist patients to dress, undress, or put on and remove supportive devices, such as braces, splints, or slings.
3| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare remains a laggard sector for physical task automation; dressing and device fitting involve vulnerable populations with diverse mobility constraints, slowing any adoption momentum. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical therapy and direct patient care remain low-digitization, high-touch environments with minimal AI/robotic adoption for hands-on tasks like this. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | While AI could theoretically assist via image-based coaching or preparation (e.g., identifying correct device placement), the core physical task requires direct human contact, limiting meaningful augmentation benefit to minor workflow steps. |
| Augmentation potential | claude-sonnet-5 | 1/5 | Current AI offers essentially no meaningful assistance for the physical act of dressing patients or fitting braces and slings. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical manipulation of clothing and adaptive devices on a patient's body, involving dexterity, force control, and real-time adjustment to patient comfort and injury status. Current AI systems lack embodied robotics capabilities for reliable, safe assistance in this context at scale. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a hands-on physical task requiring fine motor manipulation, dexterity, and physical support of patients' bodies, which current AI systems cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Patient safety, liability exposure, and the requirement for human physical contact and judgment during intimate assistance with vulnerable patients create strong legal and regulatory barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Physical contact, safety, and liability concerns around handling patients and medical devices create strong barriers, though not strictly a licensing requirement for this specific subtask. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The specialized robotics hardware, sensing, safety redundancy, and per-patient calibration required would far exceed the hourly cost of a PTA, even if a system existed today. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this physical task, so any hypothetical robotic solution would be far more costly than a human assistant given current robotics costs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI or robotic product reliably performs unsupervised dressing/undressing and orthotic device fitting on human patients in clinical settings today. This remains a research-stage capability with significant safety and precision constraints. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI or robotic product performs patient dressing or device application in clinical practice today; this remains beyond current physical automation capability. |
Perform postural drainage, percussions, or vibrations or teach deep breathing exercises to treat respiratory conditions.
3CI 0–5 · exposure 0 · augmentation 25 · importance 3.6/5 · click for rater detail
Perform postural drainage, percussions, or vibrations or teach deep breathing exercises to treat respiratory conditions.
3| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare adoption of AI remains cautious for hands-on clinical interventions, particularly in settings requiring physical contact and licensed personnel oversight. Current adoption is limited to administrative and documentation tasks, not clinical care delivery. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical therapy and hands-on healthcare delivery are among the slowest sectors for AI/robotic adoption due to physical manipulation requirements and regulatory oversight. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI might assist with educational content or video-guided breathing instruction, but the core task of performing postural drainage and percussion remains fundamentally manual and requires in-person clinical judgment. Augmentation potential is minimal for the hands-on components. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could help teach or reinforce deep breathing exercises via instructional apps or reminders, but it offers minimal assistance for the manual percussion/vibration components. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires direct physical contact, manual manipulation, and real-time assessment of patient response and tolerance. Current AI systems cannot perform hands-on postural drainage, percussions, or vibrations, and remote instruction of breathing exercises lacks the tactile feedback and individualized adjustment essential for respiratory treatment. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a hands-on manual therapy task requiring physical manipulation of a patient's body (percussion, vibration, positioning) and real-time tactile assessment, which current AI systems cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Physical therapy assistants operate under licensure and must work under the supervision of a licensed physical therapist. Direct patient contact, medical decision-making about respiratory conditions, and liability for treatment outcomes create hard regulatory and professional barriers to automation or unsupervised AI deployment. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Direct physical patient contact and clinical judgment about respiratory status typically require a licensed/certified practitioner, creating strong liability and regulatory barriers to substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The task requires in-person physical contact and continuous clinical judgment, making remote or automated delivery either impossible or unsafe. Any AI system that could theoretically assist would require expensive hardware (robotic systems) and clinical supervision, exceeding the cost of a human physical therapist assistant. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute performing this physical task, so no cost comparison favors AI; a human PTA remains the only viable option. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product performs the physical manipulation aspects of this task. While video-based breathing instruction exists, it cannot replace the hands-on assessment, adjustment, and direct intervention that define postural drainage and percussion therapy in clinical practice. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed products exist that physically perform postural drainage or chest percussion; this remains firmly in the physical/manual domain outside current AI product capability. |
Administer active or passive manual therapeutic exercises, therapeutic massage, aquatic physical therapy, or heat, light, sound, or electrical modality treatments, such as ultrasound.
0CI 0–0 · exposure 0 · augmentation 38 · importance 4.4/5 · click for rater detail
Administer active or passive manual therapeutic exercises, therapeutic massage, aquatic physical therapy, or heat, light, sound, or electrical modality treatments, such as ultrasound.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Physical therapy is a hands-on, in-person service sector with low digitization. Adoption of AI or robotics for manual treatment administration remains minimal, with the sector heavily dependent on human practitioners. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Healthcare direct-care physical therapy settings are low-digitization, hands-on environments with minimal AI/robotic adoption for actual treatment delivery, unlike documentation or scheduling functions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could marginally assist in treatment planning, patient data tracking, or guidance systems, but the core task—hands-on manual therapy—offers limited scope for meaningful AI assistance while the human remains in the loop. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-enabled devices (e.g., smart ultrasound units, sensor-based exercise tracking, biofeedback apps) can help PTAs monitor dosage, technique, and patient progress, offering moderate assistance without replacing the manual task itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires hands-on physical manipulation of a patient's body, direct tactile assessment, and real-time response to patient feedback and anatomy—capabilities that current AI cannot perform. Robotics might eventually assist, but no current general-purpose AI system can independently administer manual therapy or modality treatments. |
| Task automatability | claude-sonnet-5 | 1/5 | This task requires hands-on physical manipulation, manual palpation, and real-time tactile feedback to adjust force and technique on a patient's body, which current AI systems cannot perform. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Physical therapy assistants must be licensed or credentialed in most jurisdictions, and direct patient contact and hands-on treatment create strong legal, liability, and regulatory barriers to substitution. Patient safety and informed consent requirements are high. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Physical therapy interventions typically require a licensed practitioner or supervised assistant, with direct physical patient contact, liability for injury, and regulatory scope-of-practice rules that firmly restrict who may administer these treatments. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Manual physical therapy is a low-cost labor service; even with hypothetical robotic assistance, capital, maintenance, and liability costs would far exceed the loaded wage of a physical therapy assistant in most settings. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI-only substitute for physical manipulation of patients, so any comparison to human labor cost is moot; specialized robotic equipment (if used) is far more capital-intensive than a PTA's wage. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product performs manual therapeutic exercises, massage, or hands-on physical modality administration. While some therapeutic devices have automation, they are not general AI systems and require human operation and oversight. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs manual therapeutic exercise, massage, or hands-on modality application; robotic rehabilitation devices remain research-stage or narrow adjunct tools, not substitutes for a human assistant's hands-on work. |
Transport patients to and from treatment areas, lifting and transferring them according to positioning requirements.
0CI 0–0 · exposure 0 · augmentation 25 · importance 4.4/5 · click for rater detail
Transport patients to and from treatment areas, lifting and transferring them according to positioning requirements.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare physical therapy remains labor-intensive and resistant to automation; adoption of autonomous patient handling is negligible in production settings. Sectors employing PTAs show slow digitization and high preference for human touch and accountability. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical therapy and direct patient care are low-digitization, high-touch sectors with minimal robotic automation of manual patient handling in production use. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | Limited assistive tools exist (mechanical lifts, exoskeletons) that can reduce strain on staff, but current AI and robotics offer minimal augmentation for the core positioning and transfer judgment required. Most assistance comes from mechanical aids rather than intelligent automation. |
| Augmentation potential | claude-sonnet-5 | 2/5 | Mechanical lift devices and scheduling/logistics software can assist in planning transport, but AI itself provides little direct augmentation to the physical lifting and transferring act. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical manipulation of human bodies in variable, uncontrolled environments with significant safety and comfort considerations. Current robotics cannot reliably handle the dexterity, force modulation, and adaptive positioning needed for safe patient transfer across diverse clinical settings. |
| Task automatability | claude-sonnet-5 | 1/5 | Physically transferring and lifting patients requires embodied robotic manipulation of humans, which is far beyond current deployed AI or robotics capability; no off-the-shelf system can do this safely. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Patient safety, liability, and duty-of-care laws require human judgment and accountability; a licensed or certified human must typically oversee or directly perform patient transfers. Regulatory frameworks (OSHA, healthcare standards) and institutional risk management create hard barriers to autonomous substitution. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Patient handling involves direct physical care, safety liability, and often requires licensed/certified personnel trained in safe transfer techniques, creating strong regulatory and liability barriers to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Robotic systems capable of safe patient handling remain expensive ($100k+), require infrastructure modification, and need substantial oversight. The loaded wage of a PTA is far lower than the total cost of ownership for such systems per task instance. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI/robotic substitute deployed, so any hypothetical solution (specialized medical robots) would be far more expensive than a human PTA performing this task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | While some prototype mobile manipulators exist in research, no deployed product reliably performs full patient lifting and transfer in clinical settings at production scale. The liability and safety requirements preclude commercial deployment of autonomous systems for this task today. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No commercial product performs patient lifting/transfer autonomously in clinical settings today; this remains research-stage in assistive robotics. |
Administer traction to relieve neck or back pain, using intermittent or static traction equipment.
0CI 0–0 · exposure 0 · augmentation 25 · importance 3.9/5 · click for rater detail
Administer traction to relieve neck or back pain, using intermittent or static traction equipment.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare adoption of automation for hands-on patient care remains minimal due to regulatory requirements, liability concerns, and the need for human judgment and presence in direct patient contact. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical therapy is a hands-on, in-person healthcare sector with minimal AI-driven automation of manual treatment delivery; adoption in this specific physical task domain is essentially nonexistent. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist by recommending traction parameters or monitoring patient feedback via structured intake, but the core clinical skill—physical equipment administration and real-time patient assessment—remains fundamentally human. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could assist with documentation, treatment planning, or reminders around traction protocols, but offers negligible assistance to the actual physical administration of traction. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Administering traction requires physical handling of equipment, patient positioning, and real-time assessment of patient response and comfort. Current AI systems cannot physically manipulate equipment or safely interact with human patients in clinical settings. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires hands-on physical setup and monitoring of a patient with traction equipment, a manual clinical procedure that current AI systems cannot perform end-to-end as they lack physical embodiment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Patient safety, medical liability, and clinical scope-of-practice laws create hard barriers. A licensed healthcare provider must be responsible for patient contact, equipment setup, and monitoring during traction therapy. |
| Adoption barriers | claude-sonnet-5 | 5/5 | This is a licensed clinical procedure requiring hands-on patient assessment, equipment setup, and safety monitoring under professional scope-of-practice regulations, with direct physical contact and liability requirements. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Specialized robotic systems capable of safe patient traction would be extremely expensive to acquire, maintain, and validate compared to the loaded wage of a physical therapist assistant. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI system capable of performing this physical task, so no cost comparison favoring AI exists; a human PTA is the only viable option. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product performs clinical traction administration. This task demands embodied robotics with haptic feedback and clinical validation, which does not exist in production healthcare settings today. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product administers physical traction therapy; this remains purely a physical, human-performed clinical task with no robotic or software substitute in production. |
Perform therapeutic wound care.
0CI 0–0 · exposure 0 · augmentation 25 · importance 2.9/5 · click for rater detail
Perform therapeutic wound care.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Physical therapy remains a low-automation sector dependent on in-person evaluation and treatment; digital adoption focuses on scheduling and documentation rather than task displacement. There is minimal production-level AI substitution for wound care delivery. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical therapy and hands-on clinical care remain low-digitization, high-touch sectors with minimal AI/robotic adoption for direct physical treatment tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could marginally assist with wound imaging analysis or care plan documentation, but the core therapeutic task requires human skill and touch. Augmentation potential is limited because the task is primarily hands-on physical intervention. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with documentation, wound image analysis for tracking healing progress, or treatment protocol suggestions, but offers limited direct assistance during the physical care process itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Therapeutic wound care requires direct physical manipulation, assessment of tissue condition, sterile technique, and real-time adaptation based on patient response—all performed in-person. Current AI systems cannot perform hands-on wound management tasks. |
| Task automatability | claude-sonnet-5 | 1/5 | Wound care requires hands-on manual manipulation, debridement, dressing application, and real-time tactile assessment that current AI systems cannot physically perform. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Therapeutic wound care is restricted to licensed practitioners (physical therapists) or their direct-supervised assistants under state-specific regulations. Healthcare licensing, liability, patient safety, and the legal requirement for human clinical judgment create hard barriers to automation. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Wound care is a licensed clinical procedure requiring hands-on physical contact and accountability for patient safety, with strict regulatory and liability requirements for a credentialed provider. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | AI cannot substitute for wound care delivery itself, so cost comparison is not applicable. Any AI application would be supplementary documentation or analysis, far less valuable than the human labor required for the hands-on procedure. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this physical task, so cost comparison favors the human by default; robotic wound care remains experimental and expensive if it exists at all. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product can independently perform wound care. While AI can assist with image analysis or documentation, the core task of wound cleansing, dressing application, and therapeutic assessment demands human clinical presence. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed products autonomously perform therapeutic wound care; this remains a physical, hands-on clinical task requiring human dexterity and judgment. |
Related occupations — Healthcare Support
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.