Physical therapy · Clinical AI data

Real PT clinical records for training healthcare AI models

5,000+ de-identified physical therapy initial evaluation forms — authored by licensed DPTs — and ready for model training.

Download 5 free records View dataset details
Authored by licensed DPTs
HIPAA-compliant de-identification
Ready for model training
Full commercial license
The problem

Healthcare AI is bottlenecked by clinical data quality

Public datasets are too generic. Synthetic data lacks clinical realism. Getting real records from providers is slow, expensive, and legally complex.

Real clinical language

Written by an active DPT in authentic clinical context — not templated or synthetically generated.

Ready on day one

No procurement lag, no NDAs with health systems. Purchase, download, and begin training immediately.

Commercially licensed

Full commercial rights included. Use in proprietary models, products, and training pipelines without restriction.

5,000+
PT initial evaluations
100%
De-identified
DPT
Clinician authored
What's in the data

Physical therapy initial evaluations — in full clinical detail

Each record follows the standard PT IE structure used in outpatient musculoskeletal practice, sourced from an active clinical EMR.

01

Chief complaint & history of present illness

Patient-reported onset, mechanism, pain descriptors, prior treatment history, functional limitations, and patient goals.

02

Validated outcome measures

Standardized functional assessments including Oswestry Disability Index, Neck Disability Index, QuickDASH, and others — with full item-level scoring.

03

Objective physical examination

Range of motion, manual muscle testing, special orthopedic tests, neurological screens, joint mobility, palpation findings, and balance assessment.

04

Assessment & clinical impression

DPT-authored clinical reasoning, working diagnosis, functional deficit summary, and ICD-10 diagnosis codes.

05

Plan of care & billing data

Treatment frequency, measurable goals, intervention approach, and CPT billing codes with time and unit data.

About the data source

Clinician-authored. Not synthetic.

Our dataset is sourced from practicing Doctors of Physical Therapy with years of outpatient clinical experience, documented in WebPT — one of the most widely used EMR platforms in physical therapy. Every record reflects real clinical decision-making, real patient presentations, and real documentation patterns — then de-identified to remove all protected health information before distribution.

Start with 5 records, free

No commitment. Download a representative sample and evaluate the data quality before you buy.

Get your free sample
Enter your email — records delivered instantly