Can AI Track Your Recovery During Physiotherapy?

What can be tracked reliably

Exercise completion. Whether you did your exercises, how many repetitions and on which days. This is the most reliable data and arguably the most useful, because consistency drives recovery.

Range of motion. Camera-based apps can estimate how far a knee bends or a shoulder lifts. For large joints moving in clear view, estimates are often reasonably close to clinical measurements, though accuracy varies by product and setup.

Activity levels. Step counts, walking time and general activity from phones and wearables give a broad picture of how active someone is day to day.

Walking speed and pattern. Some systems estimate walking speed and symmetry, which are useful recovery markers after surgery or stroke.

What's harder to track

Quality of movement. Detecting compensation, such as a hip hiking to help a stiff knee, is harder for algorithms and depends heavily on camera angle.

Pain and effort. These still rely on the patient reporting them.

Why progress has stalled. Data can show a plateau. It can't tell you whether it's because of scar tissue, fear of movement, overtraining or a problem that needs medical review.

How physiotherapists use the data

When recovery data is shared, a physiotherapist can see trends between sessions: bend improving steadily, walking distance plateauing, exercises skipped on weekends. That informs decisions about progressing exercises or looking more closely at a problem. Our article on wearable devices in physiotherapy looks at the sensor side of this.

Using tracking sensibly

• choose tools your physiotherapist can see or you can share easily
• use the same setup each time: same camera position, lighting and clothing
• treat numbers as trends, not verdicts on single days
• report pain and unusual symptoms directly rather than relying on an app

Why adherence data may be the most useful

Of everything technology can track in rehab, whether exercises are actually done may be the most valuable. Many patients do far fewer home exercises than prescribed, and clinicians often don't know. When recovery stalls, it's hard to tell whether the programme is wrong or simply not happening.

Tracking changes that conversation. If a patient's knee bend has stalled and the data shows exercises done only twice in a week, the fix is about routine and barriers. If exercises are done daily and bend still stalls, the physiotherapist knows to look for another cause.

That kind of information improves decisions without needing sophisticated AI. Simple logging, reminders and shared data already make a difference.

The bottom line

AI can track some parts of recovery well, particularly exercise adherence and joint range. It works best as support for a physiotherapist's judgement, not a replacement. If something hurts, or recovery doesn't match what you expected, get assessed. For the broader picture, see our overview of AI in physiotherapy.