Independent living sensor review, 12 August to 6 September 2026

Detected bathroom-area activity was lower in the later part of the observed period. Episodes averaged about 45 a day from 12 to 24 August and about 40 a day from 25 August to 6 September, and about 34 a day across the final five days. Time with detected bathroom-area activity followed the same direction: about 205 minutes a day in the earlier window, about 181 in the later window, and about 134 a day over the final five days. The raw sensor counts agree: positive bathroom movement detections averaged about 98 a day from 12 to 24 August and about 64 a day in the final five days, although the bathroom presence sensor fell far less over the same days (about 22 to about 19 a day). Recorded movements between the two monitored areas also fell, from about 62 to about 54 a day between the halves of the period and about 47 a day in the final five days. Daily telemetry volumes over 28 August to 6 September (roughly 2,800 to 3,400 events a day) were comparable with earlier days.

The same downward direction appears whether the period is split in half or compared over shorter recent windows, which makes this the most consistent change in the data; the difference is much larger in the short windows than in the broad comparison, so how striking it looks depends on the days chosen and it cannot be said whether it goes beyond ordinary day-to-day variation. These figures describe detected room-area activity, not washing, toileting or continence, and they do not show who caused each detection. Two things temper the reading: occasions of detected presence in the bathroom held up better than the volume of movement, which may point to less movement while in the area rather than using it less often; and the movement-between-areas measure is not independent of this result, because only the kitchen and bathroom produce room-level episodes, so alternation between them falls when bathroom-area activity falls. The home was also notably cooler in the second half (bathroom average about 23.7C falling to 21.7C, peak 29.4C falling to 24.5C, with similar falls in every monitored area), which is contemporaneous context rather than an explanation. This pattern is most useful checked against what the individual and anyone supporting them say about washing and daily routine over the final fortnight.

Total detected activity episodes averaged about 113 a day over 12 to 24 August and about 110 a day over 25 August to 6 September; the final five days were lower, at about 100 a day against about 114 a day across the preceding 21 days. Looking at all positive detections by area across three successive windows (12-24 August, 25 August-1 September, 2-6 September), the bathroom and lower-stairs areas fell from about 333 to 310 to 247 a day, while the upstairs areas (top of stairs, Bedroom 1 and Bedroom 2) went from about 121 to 126 to 147 a day and kitchen-area detections stayed broadly level at about 209, 237 and 202 a day. Kitchen-area minutes were higher in the second half (about 271 against 232 a day) but the final five days were close to the earlier average, so that increase did not continue to the end of the period. Fridge door openings stayed in a similar range throughout, averaging about 33, 35 and 30 a day across the same windows, including the days when bathroom-area activity was lower. One day, 3 September, was the lowest single day (74 episodes against a 26-day average of 111, below the screening range applied to that series), yet fridge door openings that day were the highest of the 26 days.

Taken across the whole observed period, the total amount of detected activity was almost unchanged between the two halves, and detections in the upstairs areas were if anything higher at the end. The reduction in bathroom-area activity therefore coincided with activity being detected elsewhere in the home rather than with an overall drop, which is an important corrective to the finding above. The picture still has soft edges: the final upstairs window is lifted by 6 September, the highest single day for those sensors, and the last five days are genuinely lower for total episodes, so this should be treated as a direction to check rather than a settled conclusion. Counts of detections at a sensor are not counts of journeys, stairs climbed or tasks completed, and fridge door openings show continuity of a kitchen routine rather than anything about meals or nutrition.

Across 26 analysed nights, bathroom-area activity in the 22:00 to 06:00 window averaged 1.5 episodes a night. There was none at all on 15 nights, while 6 nights had three or more, the highest being 9 on the night of 17 August. Comparisons across short, intermediate and broad windows do not agree on a direction, so no clear change was identified. On the 20 nights where a value could be derived, the longest period without any detected activity averaged about 390 minutes (median 384, shortest 235, longest 478), again with no consistent direction across window sizes.

What the data shows is a wide night-to-night spread, from nights with no detected bathroom-area activity to a handful with six or nine episodes. A period without detected activity is not evidence of sleep, and detected activity is not evidence of waking or of disturbed rest, so these figures cannot describe sleep or its quality. The six busier nights are still the most useful thing here to set against any account of night-time disturbance, continence, or fluid and medication timing. Confidence is low because a quiet-period value could be derived for only 20 nights and continuous sensor availability was not confirmed for any overnight window.

The first activity detected each day between 04:00 and 10:59 averaged about 04:43 across the 26 days, ranging from 04:00 to 06:46. The two halves of the period differ by under five minutes (about 04:45 and about 04:41), and shorter recent windows are around a quarter of an hour earlier. Across the period as a whole, detected activity was concentrated between 08:00 and 21:00, with the busiest hours around 09:00 and 19:00 and very little recorded between 22:00 and 03:00.

Morning timing looks settled, which is a useful counterweight to the reduction in bathroom-area activity: the overall shape of the day held steady while the balance between areas changed. The measure only looks from 04:00 onwards and two days sit at or just above that floor, so the earliest activity on those days may have been earlier, and a first detection cannot establish when the individual woke or got up.

The first ten requested days, 1 to 10 August, contain no telemetry at all, so every comparison here rests on the 26 days with usable data from 12 August to 6 September. There is no entrance or exit sensor in the data (the only door contact is on the fridge) and no appliance telemetry, so time spent out of the property and appliance use cannot be assessed; the standard check on property exits was skipped for this reason. Room-level evidence with durations exists only for the kitchen and bathroom: the stairs, Bedroom 1 and Bedroom 2 contribute raw detections only, with no episode or duration evidence, and the lounge contributes only system connectivity data. The worn device returned no positive activity detections across the period (1,236 rows, all inactive transitions), with a further 2,484 of its rows left unclassified, so it adds no usable behavioural evidence.

These gaps shape how the rest of the report should be read. Because only two areas produce room-level evidence with durations, the counts above should not be taken to show where the individual spent most time or which rooms mattered most; lounge and bedroom use can only be glimpsed through raw detections. With no exit sensor, a quieter day inside the home cannot be distinguished from a day spent out, which remains a plausible alternative reading of the lower final days. The worn device is part of the installation but is contributing nothing usable, which is worth raising with whoever maintains the equipment before any future review relies on it.