AI In Physical Therapy: Get Used to It

I recently saw an ad campaign by ChatGPT that showed it helping with ordinary but complex human decisions such as planning road trips, suggesting stops, curating experiences. It got me thinking: if AI can manage something so human, what could it mean for us as physical therapists?

It might sound like a stretch, but to me the possibilities feel endless.

What Is AI?

Artificial Intelligence (AI) is, at its core, a system trained to recognize patterns and generate responses based on massive amounts of data. In the past five years, its growth has been staggering from clunky text generators to models capable of summarizing research, simulating reasoning, and generating custom study plans.

When I first used ChatGPT during my senior year of college, it was a novelty and good for short stories and grammar fixes. Fast forward to now: I can give it a complex neuro case, cite evidence, and it’ll synthesize plausible hypotheses, list potential differential diagnoses, and even identify discussion points I might’ve missed.

Whether you’re ready or not, it’s here to stay. You can either learn to use it, or get left behind.

How I Use It as a PT Student and Clinician

Here’s where AI already lives in my day-to-day practice:

• Brainstorming interventions: I’ll feed ChatGPT a case and say, a patient with spastic hemiplegia or vestibular dysfunction then ask for ideas, safety precautions, and progression criteria.

• Learning new topics: When I encounter a rare condition, I’ll prompt it for a concise overview before diving into deeper research.

• Building study material: It helps me generate NPTE-style questions and rationales to test my reasoning.

• Preliminary research: I use it to draft searches or summarize early evidence on things like BFR, neuroplasticity, or orthotic design.

But, and this part matters, the quality of AI’s output depends on your prompt. “BFR best evidence” gets you fluff. “What are the physiological mechanisms and evidence supporting BFR training for post-stroke hemiparesis?” gets you something useful.

Prompting is a clinical skill. The more precise your question, the more clinical reasoning the AI can help sharpen, not replace.

AI’s Current Weak Spot

AI can sound confident but still be wrong. I’ve seen ChatGPT cite papers that don’t exist, misquote data, or overstate evidence. It’s great as an initial source generator, but you must vet everything. Check every citation, verify study conclusions, and never substitute AI output for actual peer-reviewed work.

That said, there’s one bright spot I was recommended by a Medical Student friend: Open Evidence.
It’s an AI-driven database that actually connects you directly to peer-reviewed literature. Instead of guessing or generating fake references, it searches real research archives and pulls verified studies then helps summarize key findings, methods, and outcomes. I’ve found it much more efficient and accurate for pulling relevant data than traditional manual searches, especially when I need to quickly glean what the evidence says on a narrow clinical question

In short: the standard ChatGPT model is great for brainstorming and critical thinking scaffolding, but platforms like Open Evidence show what happens when AI is trained to respect the research, a glimpse into what high-quality clinical reasoning support might look like in the near future.

Where I See AI Going in PT

Let’s start with the most exciting change: documentation.
I fully expect AI to take over the bulk of charting: SOAP notes, progress reports, prior authorizations, adaptive equipment justifications. Platforms like Prompt, WebPT, and MedBridge are already working toward voice-to-text documentation systems that listen to your session and auto-generate drafts for approval.

That alone could shift clinicians’ focus back where it belongs, on the patient, not the computer.

But beyond paperwork, I think AI will evolve into an adjunct diagnostic and planning tool.
Imagine entering your findings (strength grades, tone, ROM, gait deviations) and getting AI-generated hypotheses: potential diagnoses, red flag suggestions, and progression strategies supported by research in real time.

AI won’t be palpating tissue or grading tone anytime soon. But it will help us connect dots faster, especially for clinicians who have the humility to collaborate with these systems rather than compete with them. Those who stay curious, test what AI offers, and use it as a thinking partner will sharpen their reasoning, not dull it.

As these tools refine, they’ll become integrated into EMRs, wearable data, and maybe even predictive models that alert us when a patient’s risk of falls or regression increases.

Parting Thoughts

AI won’t replace physical therapists, but PTs who use AI will be more effective than those who don’t.The earlier you start experimenting, the more natural it becomes to integrate AI into your thinking and workflow.

For beginners try using it to:

• Draft your patient education handouts.

• Create case study questions for yourself.

• Brainstorm progressions for stubborn cases.

• Translate evidence into patient-friendly language.

Get comfortable now, because this wave isn’t slowing down and the clinicians who learn to ride it will elevate outcomes, efficiency, and their own growth.

Disclaimer

I am a current Doctor of Physical Therapy (DPT) student sharing information based on my formal education and independent studies. The content presented in this newsletter is intended for informational and educational purposes only and should not be considered professional medical advice. While I strive to provide accurate and up-to-date information, my knowledge is based on my current academic and clinical rotations and ongoing learning, not extensive clinical practice.