There are 10,080 minutes in a week. A typical follow-up appointment uses somewhere between 30 and 45 of them. The other 10,000-odd minutes are where a patient actually lives, and until recently, they were almost entirely invisible to the person responsible for their recovery.
This isn't a criticism of therapists — it's a description of the tool they've been given. Standardised assessments like the Barthel Index or Lawton IADL are excellent at capturing a moment. They were never designed to capture a trajectory, because until now there was no practical way to observe one.
What a snapshot misses
Function fluctuates. An older adult recovering from a hip fracture might perform beautifully in a clinic setting — motivated, rested, supervised, on a good day — and struggle considerably more at 7pm on a Tuesday, tired, alone, without a therapist's verbal cueing. The clinic visit systematically samples the best version of a patient's performance, because it's scheduled in advance and because being observed changes behaviour. This is a well-documented distortion in almost every field that relies on episodic assessment, and rehab is no exception.
The cost of that distortion is compounding: a fall that happens at home three days after a "stable" clinic assessment doesn't get attributed to a missed deterioration, because there was no data to catch it. A gradual decline in a patient's meal-prep independence over six weeks looks, from the clinic's vantage point, like six data points that each seem fine in isolation.
What continuous monitoring actually adds
The promise of continuous, privacy-preserving home sensing isn't a firehose of raw data for a therapist to sift through — nobody has time for that, and it wouldn't be clinically useful anyway. The promise is a baseline. Once a system knows what "normal" looks like for a specific person — what time they usually start their morning routine, how long a shower typically takes them, how often they're in the kitchen — deviations become detectable in days rather than weeks.
- Trend over time-point. A single data point says almost nothing about direction; a week of data says a great deal.
- Personal baseline over population norm. "Slower than average" matters far less than "slower than they were last month."
- Early signal over late discovery. Catching a three-day drift is a phone call. Catching a six-week decline is a hospital admission.
The clinic visit isn't wrong. It's just one frame of a much longer film, and for the first time, therapists can see the rest of the reel.
This doesn't replace clinical judgment — it feeds it
None of this means an algorithm decides when a care plan changes. It means the therapist walks into a review already knowing that bathing has started forty minutes later than baseline for the third day running, and can spend the appointment asking why instead of spending it discovering that. The clinical reasoning stays exactly where it belongs — with the therapist — but it's now grounded in weeks of real-world pattern instead of forty-five minutes of performance.
The clinic visit will always matter. Hands-on assessment, rapport, the things a good therapist notices that no sensor ever will — none of that goes away. What changes is what happens between visits: instead of a gap, it becomes the part of the record that was always the most important, and, until now, the hardest to see.