Inside the Black Box: A Physical Therapist’s Experience as a Patient in AI-Driven MSK Care

Inside the Black Box: A Physical Therapist’s Experience as a Patient in AI-Driven MSK Care

When my new insurance plan offered Sword for musculoskeletal pain, I decided to enroll. Not because I was desperate for care—I’ve managed chronic neck pain for over 25 years—but because I was deeply curious.

Sword had been on my radar since around 2021, and as both a physical therapist and an entrepreneur who has delivered digital MSK care for nearly a decade, I wanted to understand what was inside the “black box” of large-scale, AI-powered rehabilitation.

I completed nine sessions of Sword’s Thrive program as a patient. What follows is my experience—filtered through the lens of 35+ years as a physical therapist, former professor, and orthopedic clinical specialist.

Before I go further, I want to be very clear about my stance.

I am not anti-digital care.
I am anti posing as physical therapy when what is delivered does not meet evidence-based standards of the profession.

Physical therapy already suffers from a branding problem. Many people don’t understand what we do, what distinguishes skilled PT care, or what “evidence-based” actually means in practice. When digital programs label themselves as physical therapy, the bar matters. A lot.

Where Sword Gets It Right: Access and Infrastructure

Sword’s strongest contribution is access.

Enrollment was easy. Within minutes of signing up, I had a telehealth intake session with a physical therapist. That kind of immediacy is impressive and meaningful—especially in a healthcare system where access to PT remains limited. Research I conducted nearly two decades ago showed that only about 7% of people with MSK conditions in the U.S. ever receive physical therapy. Any model that reduces friction deserves attention.

The technology is also well executed. The app is intuitive, messaging with clinicians is straightforward, and the tablet-based program is simple to set up and use. For some patients—particularly those comfortable with screens—this format may feel approachable and motivating.

Sword’s educational platform (Academy) is extensive and well produced, offering the type of condition-based and lifestyle education you would expect from physical therapy.

That said, technology does not automatically equal access. In the rural Appalachian community where I currently work, unreliable internet, limited cellular service, and low health literacy remain real barriers. Digital care expands access for some—but not all.

The Sword Intake Is Not a Physical Therapy Evaluation—and That Matters

My initial telehealth session lasted about 10 minutes. Most of that time was spent explaining how the program works. Very little information was gathered about me.

What occurred functioned as screening, not a physical therapy evaluation.

There was:

• Minimal history-taking

• Limited red flag screening

• No meaningful assessment of yellow flags

• No exploration of prior treatments, symptom behavior, or my goals

Without adequate data, it is impossible to design a truly individualized rehabilitation plan. This isn’t a philosophical concern—it’s a clinical one. Evaluation is the foundation of evidence-based physical therapy.

When Technology Defines the Care

As I progressed through the program, it became clear that the exercises offered were shaped less by clinical reasoning and more by what the computer vision system could monitor.

Many evidence-based exercises commonly used for neck and shoulder pain were excluded—not because they lack value, but because they cannot be easily analyzed by a front-facing camera. Postural subtleties that a human clinician can immediately detect were missed entirely by the system.

I was able to perform multiple exercises with poor posture—forward head position, thoracic kyphosis, lumbar compensation—without any corrective feedback. I even intentionally “cheated” movements to test the system and was rewarded with full scores.

The technology is not yet sensitive enough to guide or correct movement quality at the level required for rehabilitation. As a result, the care appears constrained by the limits of the tool rather than informed by clinical standards.

Under-Dosed, Under-Progressed Programming

I completed nine sessions, which felt arbitrary and insufficient—especially for chronic pain.

Exercises changed very little across sessions. When I reported no fatigue and shared that my regular routine included yoga and strength training, progression consisted mainly of increasing repetitions. No assessment of strength, endurance, or load tolerance was performed.

At one point, I was sent a PDF of weighted exercises through the messaging app, but those exercises never appeared in the program itself. The AI assistant continued guiding me through unloaded movements without acknowledging the change.

This created confusion and highlighted a deeper issue: physical therapists appeared limited in their ability to meaningfully modify care outside the AI-defined pathway.

Messaging That Misses the Mark

Language matters in pain care.

Throughout the program, I was told I was “strengthening my shoulders” while performing unloaded range-of-motion exercises and reporting zero fatigue. ROM was also described as stretching without any guidance on end-range holds.

When cervical side bending caused pain, the movement was removed entirely without explanation. For someone without pain-science literacy, this could easily reinforce fear and avoidance—an outcome we actively work to prevent in evidence-based care.

Gamification further complicated things. Visual and audio cues encouraged me to achieve five-star scores, even while being told not to push into pain. Quality of movement was secondary to hitting numeric targets.

The Missing Layer: Psychosocial Context

One of the most surprising gaps was the lack of integration of psychosocial factors.

I clearly communicated that stress influenced my neck pain and that breathing and meditation—practices from my yoga background—had been helpful. This information did not affect my program, education, or recommendations.

Sword has recently added a digital mental health program, which is encouraging. But chronic pain demands integration, not parallel silos. Physical and mental health need to talk to each other—especially in long-standing pain conditions.

Who This Program Likely Helps

Despite my critiques, this program will absolutely help some people.

It seems well suited for individuals who:

• Sit for long periods

• Have low baseline movement

• Experience mild-to-moderate, non-complex neck pain

In that context, it functions well as a movement interrupter, rather than a full rehabilitation solution.

If I Were Designing This System

I would start by collecting more data—especially during the waiting period before the device arrives. Patients are often willing to provide information if they understand how it will improve their care.

I would expand physical therapist autonomy, allowing clinicians to meaningfully alter programs rather than working around the AI.

I would create multiple recovery pathways—mobility, strength, endurance, education, stress regulation—rather than assuming one linear route to improvement.

And I would intentionally bridge physical and mental health programming for people with chronic pain.

Final Thoughts

Digital MSK care is here to stay. The real question is not whether digital care can replace in-person care, but whether it is transparent about what it is, what it is not, and who it is best designed to serve.

When digital platforms market their services as physical therapy, they assume responsibility for delivering care that reflects the full scope, rigor, and standards of the profession—not a narrowed version shaped by technological limitations.

Physical therapists should not merely oversee or monitor these systems. They must be enabled to practice within them at the top of their license, using clinical reasoning, evidence-based decision-making, and individualized progression as intended by professional standards.

Sword’s public commitment to the APTA transparency pledge represents an important step toward accountability in digital MSK care. However, my experience suggests a gap between that commitment and the delivery of full-scope, evidence-based physical therapy. This gap does not signal a failure of digital care—but rather an opportunity to elevate it by aligning technology with clinical practice, instead of constraining clinical practice to fit the technology.