Andi Assess

AI-generated draft

Independent living assessment report

Independent living sensor review: what changed, and what the data can and cannot show

Overall detected activity in the final week was very close to the level seen at the start of the period (about 96 compared with 97.5 detections per day, a difference of around 2%), so no sustained overall reduction in activity was established; the most striking-looking changes in this data set are largely bound up with two monitoring problems - a six-day interruption in movement data in late August and bathroom sensors that reported only sparsely from mid-August.

Assessment at a glance

RAG shows priority for professional review. Confidence shows the strength of the observation.

Amber Moderate confidence

Everyday activity and movement around the home

The pattern is one of a broadly maintained level of daily activity with a clear two-peak daily shape, interrupted by a period of missing and p...

Amber Moderate confidence

Use of monitored areas: kitchen and lounge

Kitchen-area activity reduced consistently on both measures (detections about 19% lower, detected time about 24% lower), while lounge detectio...

Amber Low confidence

Night-time activity and morning start times

Night-time records show longer uninterrupted periods without detected activity and later, more uniform first detections in the morning. Both p...

Grey Low confidence

Bathroom-area monitoring

The disappearance of bathroom-area detections matches a period in which the bathroom device's reporting became sparse across all of its data s...

Amber Moderate confidence

Home environment context

The home recorded a warm spell in the first half of August followed by a cooler early September. This is useful background when interpreting a...

Grey Low confidence

Areas and activities this installation cannot assess

Two areas that practitioners often expect to be covered - going out of the property and use of appliances such as a kettle or cooker - are out...

Red prompt review · Amber discuss or check · Green no notable adverse change identified · Grey not assessable or not reportable

Service-user reference: 8bf478bc-f859-4df9-a7d0-56652773e7f4_20260908_163724
Assessment period: Requested 1 August to 6 September 2026 (37 calendar days, Europe/London). Telemetry of some kind was received on all 37 days; activity-classified data was usable on 32 days, and 29 complete overnight windows contained activity evidence. No movement data was recorded between 24 and 29 August.
Generated: September 13, 2026

Executive summary

A general review of independent living sensor data for the period 1 August to 6 September 2026, undertaken without a specific referral question, to identify any patterns in daily activity, use of monitored areas and night-time activity that deserve professional attention, and to be clear about what the available evidence cannot determine.

Kitchen-area activity reduced on both available measures when comparing fully covered periods: detections fell by about 19% (roughly 38 to 31 a day) and detected time in the kitchen by about 24% (roughly 155 to 118 minutes a day). This is the one change that held at every comparison window tested and is not explained by missing data.

Lounge-area detections were about 25% higher in 31 August-6 September than in 1-22 August, while across the whole period the lounge detection series is broadly unchanged, so this is a recent rather than a sustained shift; detected time in the lounge fell by about 12% over the same comparison, so it reflects more frequent, shorter detections rather than more time spent there, and the evidence does not support a simple transfer of time from the kitchen to the lounge.

Bathroom-area detections fell from around 7 a day in early August to almost none, with virtually no bathroom detections after 24 August; the bathroom sensors reported on only 20 of the 37 days and stopped entirely after 2 September, so this cannot be read as a change in the individual's behaviour.

Three safety-device events were recorded from a worn device - fall detections on 8 August (12:41) and 21 August (09:04) and a manual alarm on 20 August (15:49); their meaning has not been confirmed.

Changes over time

Daily activity and movement between areas dipped sharply between 22 and 30 August and then returned to earlier levels; this dip coincides with the days on which no or partial movement data was recorded, so it should not be treated as a behavioural change.
First detected activity of the day became later and more uniform (about 07:25 on average in the first half of the period against about 08:33 in the second, and roughly 09:00-09:20 on each of the final seven days), and the longest nightly period without detected activity lengthened from about 5 hours to around 6 hours 20 minutes; both changes coincide with the loss of bathroom detections, which previously produced nearly all the early-morning and overnight records, and neither can indicate anything about sleep.

Overall interpretation

Taken as a whole, the data does not show a sustained reduction in everyday activity: the final week is comparable to the opening weeks. The change that survives every comparison is the reduction in kitchen-area activity, on both detections and detected time, which would be worth checking against what the individual and any family or carers describe about meals and daily routine. The most important limitation is that sensor continuity has not been independently verified: bathroom monitoring became sparse from around 11 August and all movement data was absent for six days in late August, so bathroom-area, overnight and morning-timing results cannot be separated from sensor availability, and the evidence cannot determine whether the bathroom change reflects a fault or a change in use. The next practical steps are to establish what happened to the bathroom sensors and to the late-August data, and to obtain a further period of complete data before drawing any conclusions about night-time or personal-care routines.

Assessment finding

Everyday activity and movement around the home

Amber Moderate confidence

What was detected

Across days with usable data, an average of about 91 activity detections and about 48 changes between monitored areas were recorded per day. Movement between areas was at a similar level at the end of the period as before it: about 47 changes a day across 4-6 September, against an average of about 48 a day for all the preceding days with data.
Comparing fully covered periods, daily detections averaged 97.5 between 1 and 22 August and 96.0 between 31 August and 6 September, a difference of about 2%; the standard summary figure of a 28% reduction is based on periods that include the days affected by the monitoring interruption.
Across the period, detected activity concentrated in the late morning (around 271 detections in the 09:00 hour and 263 in the 10:00 hour) and the early evening (around 274 in the 18:00 hour, 290 in the 19:00 hour and 213 in the 20:00 hour), with very little detected between midnight and 08:00.

Pattern identified

The pattern is one of a broadly maintained level of daily activity with a clear two-peak daily shape, interrupted by a period of missing and partial data in late August. Because the apparent overall decline depends entirely on which dates are included, no clear change in overall activity was identified; this matters because a headline reduction could otherwise be mistaken for deterioration.

Why this may matter

Any comparison that spans 23-30 August will understate activity because of missing movement data rather than because of the individual's behaviour.
A longest daytime gap of about 3 hours between detections was recorded; periods without detections do not establish that nothing was happening.
Detection counts depend on sensor placement and sensitivity as well as on movement.

What to explore

Was anything different about the period 23-30 August, for example a service visit, a repair, a power interruption or time away from home?
Does the two-peak daily pattern match what the individual describes as their usual day?

What the data cannot tell us Sensor installation continuity and like-for-like coverage across comparison periods have not been independently verified., Absence of a detected event does not prove that an activity did not occur., and Counts of detections cannot show what the individual was doing, how well they were managing, or whether anyone else was present.

Assessment finding

Use of monitored areas: kitchen and lounge

Amber Moderate confidence

What was detected

Kitchen-area detections fell from an average of about 38 per day (1-22 August) to about 31 per day (31 August-6 September), around 19% lower, and detected kitchen time fell over the same windows from about 155 to about 118 minutes a day, around 24% lower. The same direction of change appeared at every comparison window tested.
Lounge-area detections rose over the same windows from about 39 to about 48 per day (around 25% higher), while detected lounge time fell from about 290 to about 256 minutes a day (around 12% lower); across the whole period the lounge detection series is broadly unchanged, so the rise is a recent local change rather than a sustained one.
Across the whole period, detected activity was spread between the lounge (42% of room detections), kitchen (37%), hallway (18%) and bathroom (3%), with average detected occupied time of about 262 minutes a day in the lounge and 137 minutes in the kitchen.

Pattern identified

Kitchen-area activity reduced consistently on both measures (detections about 19% lower, detected time about 24% lower), while lounge detections rose without any matching rise in detected lounge time. Total daily activity was almost unchanged, so this is better described as a change in how activity is distributed and paced around the home than as an overall reduction, and the evidence does not support a simple transfer of time from the kitchen to the lounge.

Why this may matter

Reduced kitchen-area activity may be explored alongside information about meal preparation, shopping, appetite, use of ready-prepared food or support from others, without assuming any of these.
More frequent, shorter lounge detections can arise from many causes, including movement patterns, seating position and sensor sensitivity; the comparison window for the rise is only seven days.
Detected activity in an area is not proof of a visit, of a task completed there, or of the amount of time spent there.

What to explore

Has anything changed recently about how meals and drinks are prepared or provided?
Does the individual describe any change in how they move around or settle in the lounge during the day?

What the data cannot tell us The recent comparison relies on a seven-day window, so it may not persist; it should be re-checked over a longer period of complete data., The evidence cannot establish why detections in an area changed, and appliance use could not be derived to corroborate kitchen activity., and Sensor continuity for the kitchen and lounge streams has not been independently confirmed.

Assessment finding

Night-time activity and morning start times

Amber Low confidence

What was detected

The longest nightly period without detected activity averaged about 5 hours 35 minutes across 29 analysed nights and lengthened consistently, from around 5 hours in early August to about 6 hours 20 minutes on each of the most recent nights; this describes detected activity only.
Overnight bathroom-area detections averaged about 2 per night to 10 August and none on the most recent nights; 17 of the 29 analysed nights had no overnight bathroom-area detection at all, and the nights without any coincide with the period when the bathroom sensors were not reporting reliably.
First detected activity of the day became later at every comparison scale (about 07:25 on average in the first half of the period against about 08:33 in the second) and settled at roughly 09:00-09:20 on each of the final seven days; across the 30 days with a morning value, first detections ranged from 04:08 to 09:45, showing wide day-to-day variation in the earlier weeks compared with the very consistent late-morning pattern at the end.
Event-level checking shows that nearly every first detection before 07:00 was recorded by the bathroom sensors that later stopped reporting, and that from mid-August onwards the first detection of the day was recorded in the hallway or kitchen at around 09:00.

Pattern identified

Night-time records show longer uninterrupted periods without detected activity and later, more uniform first detections in the morning. Both patterns began as bathroom sensor reporting became sparse, and the early-morning detections that previously produced the earlier start times came from those same sensors. The arithmetic change is consistent across every window tested but is not safely interpretable as a change in the individual's night-time or morning routine, and it cannot indicate sleep.

Why this may matter

These figures describe detected activity and periods without detected activity only; they cannot indicate sleep, waking, sleep quality or night-time disturbance.
If the bathroom sensors are restored, the morning and overnight measures should be re-examined before any conclusion is drawn about changes in routine.
Wide day-to-day variation in the earlier weeks may partly reflect which sensors were reporting on which days.

What to explore

What time does the individual describe getting up, and does this match the late-morning first detections seen in the final week?
Are there any reported concerns about night-time comfort, continence or getting to the bathroom that other information sources could address?

What the data cannot tell us The evidence can describe overnight activity and quiet periods but cannot confirm sleep, waking, sleep quality or restorative sleep., The assessment contains 36 complete local 22:00-06:00 windows and 2 partial boundary windows; activity was observed in 29 complete windows and continuous sensor availability was confirmed in none of the 38 intersecting overnight windows., and Because bathroom monitoring was intermittent, changes in overnight and morning measures cannot be separated from sensor availability.

Assessment finding

Bathroom-area monitoring

Grey Low confidence

What was detected

Bathroom movement and presence sensors produced events on only 20 of the assessment dates and last produced any event on 2 September, while hallway, kitchen and lounge sensors produced events on 31 dates up to 6 September.
Day-by-day counts show regular bathroom detections from 1 to 11 August (between 8 and 32 positive detections a day), a smaller number on scattered days between 16 and 23 August (2 to 10 a day), and then virtually nothing: none on 24 August, two detections on 31 August and none thereafter. This is equivalent to a fall from around 7 detected bathroom-area episodes a day to almost none, and cannot be read as a change in the individual's behaviour.
On several days the same bathroom device still sent light and temperature readings while recording no movement or presence detections, and its overall reporting volume fell from several hundred readings a day in early August to a few dozen later in the month.

Pattern identified

The disappearance of bathroom-area detections matches a period in which the bathroom device's reporting became sparse across all of its data streams and then ceased altogether. Because the device was still sending some readings on days with no detections, the loss of bathroom detections cannot be attributed with confidence either to a fault or to a change in how the room is used; no conclusion about personal care, continence or bathroom use should be drawn from this period.

Why this may matter

A technical check of the bathroom sensor (power, battery, positioning, connectivity) would establish whether the loss of data is a fault.
If bathroom information is important to the assessment, a further monitoring period after any repair would be needed.

What to explore

Can the installer or provider confirm whether the bathroom sensor was faulty, moved, covered or removed during August?
Is there any other information about bathroom use during this period that could fill the gap?

What the data cannot tell us The available evidence cannot determine whether bathroom-area activity changed, because the sensors reported only sparsely for most of the second half of the period. and Absence of a detected event does not prove that an activity did not occur.

Assessment finding

Home environment context

Amber Moderate confidence

What was detected

Indoor ambient temperature readings reached 30C or above on 8 days between 1 and 14 August, with a highest recorded reading of about 33C; daily maximum readings in the lounge, kitchen and hallway were around 30C on 13 and 14 August, and readings fell to around 21C by early September.

Pattern identified

The home recorded a warm spell in the first half of August followed by a cooler early September. This is useful background when interpreting activity in that period and may be worth discussing in relation to comfort, hydration and ventilation, without assuming any effect on the individual.

Why this may matter

Temperature readings reflect the sensor's position in the room and are not a calibrated measure of the temperature the individual experienced.
Discussion of heat in the home, ventilation and hydration may be useful for planning ahead of future warm weather.

What to explore

Did the individual find the home uncomfortably warm in early and mid-August, and do they have workable ways to keep cool?

What the data cannot tell us The evidence describes recorded room temperatures only and cannot establish comfort, health effects or any link to activity levels. and Temperature reporting is uneven between rooms and was sparse in the bathroom.

Assessment finding

Areas and activities this installation cannot assess

Grey Low confidence

What was detected

Leaving and returning to the property cannot be assessed: the plausibility check for property exits was skipped because no confirmed front-door or exit telemetry was available, and hallway activity alone cannot show that someone left the home.
One appliance-classified source stream is present but usage sessions were not derived, so nothing should be inferred about appliance use, and this must not be read as no appliance use.

Pattern identified

Two areas that practitioners often expect to be covered - going out of the property and use of appliances such as a kettle or cooker - are outside what this installation can currently evidence. This limits what can be said about community access and about kitchen tasks, which is relevant when interpreting the reduction in kitchen-area activity.

Why this may matter

If getting out of the home or appliance use is material to the assessment, additional or reconfigured telemetry would be needed.
Information about outings and appliance use should be sought from the individual, family or care records instead.

What to explore

Would door or appliance monitoring add value for this individual, and would they consent to it?

What the data cannot tell us No exit or absence evidence is available, so periods away from home cannot be identified or excluded. and Appliance readings present in the data cannot support any statement about use.

Potential anomalies

Patterns for closer review

Movement data stopped entirely for six days: only 2 activity events were recorded on 24 August and none from 25 to 29 August, with reduced counts on 23 and 30 August, while several hundred other telemetry readings from the same property continued each of those days. The 23rd (40 detections) and 24th (none) are flagged as falling outside the expected daily range for this period, and a low-battery indication was recorded from a movable device on 28 August.

Continued environmental readings during the same days suggest the property's equipment remained partly online while the movement streams did not report.
A period away from home, a technical or power problem, or a maintenance change could each produce this pattern; the data cannot distinguish between them, and the meaning of the low-battery indication has not been confirmed.

Question for review Can the provider or the individual account for the six-day interruption in movement data between 24 and 29 August, including whether any device needed a battery change?

Three safety-device events were recorded from a worn device: fall detections on 8 August at 12:41 and 21 August at 09:04, and a manual alarm on 20 August at 15:49.

The meaning of these events has not been confirmed; they may include test activations, accidental triggers or genuine incidents, and they must not be treated as ordinary activity.
Any alarm-handling records held by the monitoring provider would be the appropriate source for what happened at these times.

Question for review What do the monitoring provider's response records show for 8, 20 and 21 August, and was any follow-up undertaken?

Review and limitations

Professional review

Questions for professional review

Can the bathroom sensors and the six-day movement-data interruption be checked technically, and can a further period of complete data be obtained before conclusions are drawn about night-time or personal-care routines?
Does the individual, or anyone supporting them, describe any recent change in how meals and drinks are prepared that might sit alongside the reduction in kitchen-area activity?
What time does the individual say they usually start their day, and does the consistent late-morning pattern of the final week fit with that?
What is known about the three worn-device safety events in August, and is any follow-up outstanding?

Quality warnings

Average property exits per behavioural day could not be checked because the evidence metric is unavailable.
1 SOS or safety-related telemetry events are present. Their meaning has not been confirmed and they must not be treated as ordinary activity.
1 appliance-classified source stream contains 3 readings, but usage sessions were not derived because approved activity-state and sessionisation policies are not yet available. Do not interpret this as zero appliance use.

Data limitations

Passive telemetry describes detected events and behavioural patterns; it cannot establish diagnosis, subjective experience, intention or causation.
The evidence can describe overnight activity and quiet periods but cannot confirm sleep, waking, sleep quality or restorative sleep.
Absence of a detected event does not prove that an activity did not occur.
The assessment contains 36 complete local 22:00–06:00 calendar windows and 2 partial boundary windows. Activity was observed in 29 complete windows. Continuous sensor availability was confirmed in 0 of 38 intersecting overnight windows.
Sensor installation continuity and like-for-like coverage across comparison periods have not yet been independently verified.

This AI-generated draft supports professional judgement and requires human review. It describes patterns in independent living sensor data only; it does not diagnose, assess risk, determine eligibility or recommend changes to care.