Bedside observation without a camera
A record of the night, even when the patient says nothing
Not an alarm device — a handover device. It uses no visible-light camera, so no footage anyone could be recognised in is ever made.
22:10–06:29 · 8 channels · coverage 100%Sample data
23:0000:0001:0002:0003:0004:0005:0006:00
01:40 Skin temperature starts rising02:15 18 coughs within 25 minutes03:12 One loud impact04:50 Bed empty 15 minutes
How it observes
Thermal
Body heat gives movement and skin-surface temperature. It works with the lights off, and a face cannot be made out.
Radar
A separate check on whether someone is there, on an axis that has nothing to do with heat.
Microphone
It tells coughs, snoring and vocalisation apart. No audio file is kept.
The three check one another, and where they disagree the stretch is left as “not confirmed”. More on the technology →
CareLog is not a medical device. It does not diagnose, and it does not alarm. Instead it records what was observed and what could not be observed as two separate things.
At night there is no observer
If the patient does not call and the nurse does not write, that night leaves no record. Yet the patients who most need watching are the ones least able to say what is wrong.
13.5
patients per nurse on a night shift (8.6 by day, 10.9 in the evening)
Source: Ministry of Health and Welfare and NHIS, “Survey of health workforce duties”, surveyed Jan–Mar 2023 across 84 hospitals. General wards on three shifts — not long-term-care hospitals or nursing homes. ⚠ Cited at second hand; we could not obtain the report itself, so we do not know whether this is a mean or a median, or how it splits by hospital type.
990,000
additional care workers Korea would need by 2043 to hold the caring load at its 2023 level
Source: Kwon Jeonghyun, KDI, “Projection and policy direction for the elderly-care workforce”, 16 Apr 2026. A conditional projection, not a count of today's shortfall.
“Not seen” and “did not happen” are different
To write the single line “nothing of note observed”, you need grounds for having watched the whole night without a gap. Stretches where the record breaks are kept as not confirmed, rather than did not happen.
A night confirmed throughout
22:10Recording started
22:10–23:05Nothing unusual observed by the device
~23:05Fell asleep (estimated)
23:05–23:40Nothing unusual observed by the device
23:40–23:488 coughs in a row
23:48–00:35Nothing unusual observed by the device
00:35Out of bed after increased movement (estimated)
00:43–01:50Nothing unusual observed by the device
01:50–02:02Speech 3 min
02:15–02:4018 coughs in a row
02:58–03:06Presence not confirmed
03:12A loud impact sound — the cause is not ruled on
03:40Out of bed after increased movement (estimated)
03:46–04:50Nothing unusual observed by the device
04:50Out of bed (estimated)
05:05–06:14Nothing unusual observed by the device
~06:14Woke (estimated)
06:30Recording ended
Coverage 100% · observed 8h 20m · missing 0m
A night with a stretch that could not be confirmed
22:10Recording started
22:10–23:05Nothing unusual observed by the device
~23:05Fell asleep (estimated)
23:05–00:00Nothing unusual observed by the device
00:00–00:40Nothing unusual observed by the device
00:40–03:20Presence picked up by radar only (body not confirmed on thermal)
03:20–04:00Nothing unusual observed by the device
04:00–06:00Nothing unusual observed by the device
06:00–06:30Nothing unusual observed by the device
06:30Recording ended
Coverage 68% · observed 8h 20m · body not confirmed 2h 40m
Built to be usable on a ward
There is no visible-light camera
No video or audio that anyone could be identified from is ever made. Only figures and event records are sent to the server.
It does not rule on anything
No diagnosis and no alarm — observation, and a suggestion to check, and no further.
Setting it up is simple
It sits beside the bed. The patient wears nothing, and no wiring work is needed.