Draft

A smartwatch and mobile app that monitors heart rate in real time and alerts emergency contacts the moment something goes wrong

Built specifically for elderly users — a React Native app paired with smartwatch integration that continuously monitors heart rate, detects dangerous anomalies against standard medical thresholds, and instantly sends emergency alerts to pre-set contacts with real-time location tracking. No button to press. No action required. Just continuous protection.

Months from brief to deployment

24/7

Continuous heart rate monitoring via smartwatch

Platforms — mobile app and smartwatch — working as one system

Challenge

The person who needs help most is the least likely to be able to ask for it

Emergency alert apps typically require the user to take action — press a button, shake the phone, speak a command. For elderly users experiencing a cardiac event, that assumption breaks down completely. The brief was to build a system that required nothing from the user once set up — continuous passive monitoring, automatic anomaly detection, and instant alerts to the people who matter, with location so they can actually find them. Every design and technical decision flowed from that single requirement.

C-01

No user action in an emergency

The entire alert system had to be passive — no button, no gesture, no voice command. Detection had to be automatic and alerts had to fire without the user doing anything at all. For an elderly person experiencing a cardiac event, anything that required conscious action was a design failure.

C-02

Accurate detection without false alarms

A system that cried wolf would be abandoned within days. Standard medical heart rate thresholds — bradycardia, tachycardia, no pulse detected — had to be applied correctly, with enough intelligence to distinguish a genuine emergency from momentary sensor noise or physical activity, without introducing dangerous delays in real emergencies.

C-03

Cross-platform watch compatibility

Elderly users own a wide range of smartwatches — Apple Watch, Wear OS, and others. The monitoring system had to work reliably across all of them, reading heart rate data consistently regardless of the underlying watch platform, without requiring the user to own any specific hardware.

C-04

Real-time location for contacts who need to respond

Knowing someone needs help is only useful if you can find them. Emergency contacts needed real-time location tracking the moment an alert fired — not a last-known position, not a manual check-in, but a live map they could follow to reach the person in distress as quickly as possible.

How it works

Set up once. Protected continuously. Alert fires automatically.

STAGE — 01 / DETECT

User sets up

Installs app, pairs smartwatch, adds emergency contacts

Monitoring begins

Smartwatch continuously reads heart rate in real time

Data processed

Google AI analyses heart rate stream against standard medical thresholds

Anomaly detected

Bradycardia, tachycardia, or no pulse triggers emergency protocol

Alert fires

All emergency contacts receive instant notification with alert type

Location shared

Contacts with the app installed get real-time live location tracking

Technical tension

Continuous monitoring across every watch platform — accurate enough to trust, fast enough to matter

Reading heart rate data reliably across Apple Watch, Wear OS, and other smartwatch platforms required building a unified data layer that abstracted the differences between each platform's health APIs without losing consistency or introducing latency. Heart rate data arrives at different intervals, in different formats, with different reliability characteristics depending on the watch. The monitoring system needed to handle all of that transparently — presenting a consistent stream to the anomaly detection layer regardless of which watch the user happened to own. Getting this right was the most sustained engineering challenge on the project, because any inconsistency at the data layer would propagate directly into detection accuracy. Google AI was integrated to distinguish genuine cardiac anomalies from false positives — movement artefacts, sensor dropout, and brief spikes during physical activity all had to be correctly excluded from triggering emergency alerts, while genuine bradycardia, tachycardia, and absent pulse events had to be caught without delay. The location tracking system was built to activate the moment an alert fired — sharing real-time position with emergency contacts who had the app installed, updating continuously, so responders were always navigating to where the person was rather than where they had been.

heart-rate-monitor — anomaly detection

01// continuous heart rate analysis

02const analyseHeartRate = async (bpm, userId) => {

03 // standard medical thresholds

04 const BRADY_THRESHOLD = 60 // bradycardia below 60 BPM

05const TACHY_THRESHOLD = 100 // tachycardia above 100 BPM

06// pass to Google AI for false positive filtering

07const isGenuine = await googleAI.validateReading(bpm, userId)

08 if (!isGenuine) return // movement artefact — ignore

09 if (bpm < BRADY_THRESHOLD || bpm > TACHY_THRESHOLD || bpm === 0) {

10 const alertType = bpm === 0 ? 'NO_PULSE' :

11 bpm < BRADY_THRESHOLD ? 'BRADYCARDIA' : 'TACHYCARDIA'

12 → emergency alert fired · all contacts notified

13→ real-time location sharing activated

14}

15}

Kicker-THE PROTECTION LOOP

Continuous monitoring that never stops — alert fires automatically when it matters

From the moment the watch is paired, the same loop runs continuously in the background — reading, analysing, and standing by to alert. The user never has to think about it.

Capabilities

Four things the system does — built around one principle: the user should never have to do anything in an emergency

HEART RATE MONITORING

Continuous passive monitoring via smartwatch

Real-time heart rate data read continuously from the paired smartwatch — Apple Watch, Wear OS, or any compatible platform. No user interaction required after initial setup. Monitoring runs in the background at all times, regardless of what the user is doing.

ANOMALY DETECTION

AI-powered detection against medical thresholds

Google AI analyses the heart rate stream continuously against standard medical thresholds — bradycardia, tachycardia, absent pulse. Intelligent filtering distinguishes genuine cardiac events from movement artefacts and sensor noise, minimising false alarms without delaying real ones.

EMERGENCY ALERTS

Instant alerts to every emergency contact

When an anomaly is confirmed, all pre-set emergency contacts receive an instant alert — regardless of whether they have the app installed. Alert type included so contacts know immediately what kind of event has been detected and how urgently to respond.

REAL-TIME LOCATION

Live location tracking for contacts who have the app

Emergency contacts with the app installed receive real-time live location tracking the moment an alert fires — a continuously updating map showing exactly where the person in distress is located, so they can navigate directly to them rather than relying on a last-known position.

Results

Deployed in three months. Continuous protection. No user action required in an emergency.

The full system — mobile app, smartwatch integration across all major platforms, Google AI anomaly detection, emergency alert pipeline, and real-time location tracking — was delivered and deployed in three months. The cross-platform watch abstraction layer handled heart rate data consistently across Apple Watch and Wear OS without requiring users to own specific hardware. The Google AI detection layer correctly identified standard medical threshold breaches while filtering movement artefacts and sensor noise from triggering false alerts. Emergency contacts received instant alerts with event type on anomaly detection, and contacts with the app installed received real-time location tracking immediately. The system required zero user action after initial setup — operating continuously in the background for the duration of the user's daily life.

3mo

Full system delivered — mobile, watch, AI, and location

24/7

Continuous passive monitoring with no user action required

Actions required from user during an emergency

All

Major smartwatch platforms supported from launch

Building a health or safety app where the user can't be relied on to take action?

We've built passive monitoring systems that detect, decide, and alert automatically — across mobile and wearable platforms. If your product needs to work when the user can't, that's exactly the kind of engineering challenge we take on. Let's talk.