Journal

Passive Elderly Health Monitoring: Radar Vital Signs, Smart Toilet, and Behavior Analytics

ระบบติดตามสุขภาพผู้สูงอายุแบบ Passive: Radar Vital Signs, Smart Toilet และ Behavior Analytics

May 12, 2026 · 1 min read
Passive Elderly Health Monitoring: Radar Vital Signs, Smart Toilet, and Behavior Analytics

What Is Passive Monitoring and Why It Matters for Thai Elderly

Most elderly Thais refuse to wear smartwatches or health trackers — they find them uncomfortable or feel surveilled. Passive monitoring measures health without any wearables, using environment sensors instead. The elderly person lives normally while the system works invisibly in the background.

60GHz mmWave Radar: Contactless Vital Signs

Seeed Studio MR60BHA1 (60GHz): detects respiratory rate, heart rate, and falls without any wearable device. Detection radius 3–5 metres; connects to Home Assistant via UART-to-USB or ESPHome on ESP32.

Measurements: respiratory rate 12–20 breaths/min normal (alert if <10 or >25); heart rate 60–100 bpm normal (alert immediately if >120); presence detection — detects whether someone is in the room even without movement, unlike PIR which only detects motion.

Installation: mount on ceiling or wall above the bed, 1–2 metres from the person. Ceiling centre of bedroom is optimal.

Smart Toilet: Bathroom Visit Frequency Tracking

Toilet visit frequency is an important biomarker. Elderly adults normally visit 5–8 times per day. An increase to >10 visits per day may indicate UTI (urinary tract infection), diabetes, or medication side effects.

Tracking method: PIR or mmWave radar in the bathroom logs a timestamp each time someone enters. Home Assistant counts daily visits and compares to a personal baseline (calculated over the first 7 days of monitoring). If visit count rises >50% above baseline, the family is alerted.

Kitchen/Fridge Access Pattern Analytics

A magnetic contact sensor on the fridge door logs every opening — normally 4–8 times per day around meal times. If the fridge is not opened for >6 hours during a period when the person normally eats, this may indicate illness. Home Assistant builds a personal Pattern Profile from historical data.

Withings Body+ Smart Scale: BIA Body Composition

Daily morning weigh-in: weight, BMI, body fat %, muscle mass, hydration %. Data syncs from the Withings Health Mate API → Home Assistant. If weight drops >2kg in 7 days → alert (possible dehydration or illness).

IoT Pill Dispenser

An automatic pill dispenser (e.g., MedMinder or Hero Health) holds pre-loaded medication for multiple doses. It dispenses at scheduled times and sends a LINE notification. If the pill is not taken within 30 minutes, a second alert is sent. After 1 hour without retrieval, the family is notified.

ML Anomaly Detection for Behavioural Patterns

Home Assistant + Python analyses a multi-dimensional behaviour vector daily: wake-up time (from bed-exit radar), first bathroom visit time, bathroom visit frequency, first fridge opening time, mean overnight heart rate, and morning weight. An Isolation Forest algorithm detects statistically significant anomalies. On anomaly days, a summary report is sent via LINE OA to the family at 07:00 with a traffic-light status: Green (normal), Yellow (worth observing), Red (call to check in).

PDPA Compliance

All sensors operate locally — no data is sent to third-party cloud services. Vital signs and behaviour pattern data are stored only in the local Home Assistant database. The elderly person and family retain full control of their data.

Questions & answers

How accurate is mmWave radar for heart rate without a wearable?
The Seeed MR60BHA1 measures heart rate to ±5 bpm and respiratory rate to ±1 breath/min within 0.5–3 metres. Accuracy decreases when the person moves significantly — it is most accurate during sleep.
What does >10 bathroom visits per day indicate in elderly people?
It may indicate UTI (urinary tract infection, common in elderly women), poorly controlled diabetes, diuretic medication, or overactive bladder. Inform a doctor if the increase is sudden.
What is Isolation Forest and how does it work in Home Assistant?
Isolation Forest is an ML algorithm that detects anomalies based on the principle that abnormal data points are easier to isolate than normal ones. Using Python (sklearn) on HA, it trains on 30 days of behaviour data then scores each day — a low score triggers an anomaly alert.
Does this system require internet, or can it run entirely locally?
Mostly local. The Withings scale needs brief internet access to sync via the Withings API, and LINE notifications require connectivity. Vital sign radar, behaviour analytics, and ML anomaly detection are all 100% local.

Related reading