Senior
Real-time eldercare monitoring, from wearable to ambulance
Overview
A monitoring platform for seniors living independently, turning signals from wearable devices into a verified human response, fast enough to matter.
Seniorsoft came to us with a hard problem: an ageing parent living alone is fine until the moment they are not, and the gap between an incident and someone finding out is measured in hours. Consumer wearables detect falls, but a detection that reaches nobody is not a safety system. What was missing was everything between the sensor and the ambulance.
We built this platform around that gap. A fleet of connected devices (pendants, wrist bands, watches and automated pill dispensers) streams continuously into a telematics pipeline. Events are classified into the categories that actually change what happens next: an SOS press, a detected fall, a geo-fence breach, a missed medication window, or a wellness anomaly.
Crucially, the system does not auto-dispatch. Every alert lands in a monitoring console where a human operator verifies it before escalating, because a false fall alert that sends an ambulance costs a family money and trust. Operators can acknowledge, call the senior, notify family, escalate to an administrator, or dispatch, with each action timestamped against a ticket with response times tracked.
Around that core sit the operational systems a real service needs: role-scoped portals for administrators, guardians and monitoring staff; a device registry with IMEI lookup, assignment and battery health; location management; and ERP integration so billing reflects what was actually delivered. Sign-in supports both password and mobile OTP, because the people who use it are not all comfortable with either alone.
About
Signal to response.
A connected-health platform that turns wearable telemetry into a verified human response, with the escalation path, audit trail and billing that a real monitoring service depends on.
Challenge
Seniors living independently are safe right up until they are not, and the delay between an incident and anyone noticing is the whole problem. Off-the-shelf wearables raise alerts into a void: no operator, no verification, no escalation path, no record. Meanwhile a monitoring service that dispatches on every raw sensor event burns through ambulances and family goodwill on false positives: a dropped pendant reads much like a fall. The hard part was never detection. It was everything between the signal and a person who can act on it.
Goal
Build the operational layer around the devices: continuous ingestion from a mixed fleet of pendants, wrist bands, watches and pill dispensers; classification into event types that map to distinct responses; a monitoring console where trained operators verify before escalating; and role-scoped access so administrators, guardians and monitoring staff each see exactly what their job requires, and nothing more.
Solution
We built a real-time platform with a human deliberately kept in the loop. Device telemetry streams through a telematics gateway into AWS, where events are classified as SOS, fall, geo-fence breach, medication or wellness. Each becomes a ticket in an operator console with the senior’s last known location, device battery and alert history attached. Operators acknowledge, then choose: call the senior, alert the family, escalate to an administrator, or dispatch an ambulance, with every step timestamped so response times are measurable rather than assumed. Supporting it: a device registry with IMEI lookup and assignment tracking, location management, dual password and mobile-OTP authentication, and ERP integration so billing follows real delivered service.
Architecture
From wearable to ambulance.
Six stages, one of which is deliberately not automated. Keeping a trained operator between detection and dispatch is what separates a monitoring service from an alarm that cries wolf.
- 01
Device fleet
Pendants, wrist bands, watches and automated pill dispensers, each tracked by IMEI with battery health
- 02
Telemetry ingestion
Continuous streams over 4G through a telematics gateway into AWS
- 03
Event classification
SOS, fall, geo-fence breach, medication window, wellness anomaly
- 04
Human verification
An operator confirms before anything escalates. No automatic dispatch on a raw sensor event
Human in the loop
- 05
Response
Call the senior, alert family, escalate to admin, or dispatch an ambulance, with every action timestamped
- 06
ERP & billing
Resolved tickets and response times flow through to billing, so invoices reflect delivered service
The build
Two codebases, one platform.
The technologies are named because they are ours to name. The shape of Seniorsoft's system is theirs, so it is not described here.
Backend
A REST API on AWS
Device events arrive in batches from a telemetry gateway, are routed by event type, and become tickets an operator can work. Open tickets stream to the console over server-sent events rather than being polled for.
- Java 21
- Spring Boot 3.5
- Spring Security
- JWT
- PostgreSQL
- Flyway
- Project Reactor
- AWS EC2
- AWS SES
- AWS SNS
- AWS Secrets Manager
Console
A single-page operator console
A console for the monitoring team, with the live alert stream held open on every authenticated screen rather than mounted per page, so an alert reaches an operator wherever they happen to be working.
- TypeScript
- Server-sent events
- Vitest
- Vercel
- GitHub Actions
This platform had devices, operators, tickets and billing sitting in separate systems, and a response time nobody could measure. If your operation is spread the same way, a discovery call is where we work out what it would take to pull it together.
Contact