Real time platform

Acute care and ICU telemetry monitoring SaaS

A hospital platform that gathers vitals and medical device telemetry in one live view and warns attending staff early when a patient's trend moves the wrong way.

Our BAA offer

Built forHospitals and acute care

Team shapeAround 8 people: architect, backend and frontend engineers, QA, DevOps, clinical analyst

TimelineAbout 9 to 10 months including load and failover testing

The problem

ICU data arrives continuously from many devices. A dashboard that polls every few seconds falls behind, and an alert that arrives late or goes to the wrong person is worse than no alert.

How we approached it

We treated the system as a streaming pipeline first and a web app second. Device readings enter a queue, are normalized by workers, evaluated by anomaly detection, then pushed to dashboards and phones over WebSockets. Alerts follow a role based routing table so the right physician or nurse is notified, with escalation if nobody acknowledges.

What we built

High concurrency WebSocket delivery of vitals and device telemetry to web dashboards
Redis backed queues with Supervisor managed workers for normalization and evaluation
Anomaly detection over continuous vital streams to flag early signs of decline
Role based alert routing and escalation by web dashboard and mobile push
Bedside and unit level views for charge nurses and attending physicians
Immutable audit logging and encrypted storage for regulatory review

Architecture

How the pieces fit together

1Devices and gateways

2Queue and workers

3Anomaly rules and alert router

4Dashboards and mobile push

IngestionDevice gateways post readings to an authenticated endpoint that writes to a queue and returns immediately.

ProcessingSupervisor keeps worker pools alive. Workers normalize units, validate ranges and run the anomaly rules, with idempotent handling so retries never duplicate readings.

DeliveryWebSocket channels are scoped per unit and per role. Reconnect logic replays missed readings from a short buffer.

AlertingA routing table maps severity and unit to people. Unacknowledged alerts escalate on a timer.

Safeguards built in

Security and compliance controls

Immutable audit log
Encrypted storage and backups
Role scoped channels
Failover tested workers
Documented downtime procedure

What changed

Outcomes for the operator

Clinicians see a patient's trend on one screen instead of walking between devices.
Alerts reach a named role quickly and escalate on their own when ignored.
The hospital can show a clear record of who saw which alert and when.

Technology

LaravelRedisSupervisorWebSocketsMySQLTime series storagePush notifications

Integrations

Device gatewaysHospital paging and mobile pushADT feeds for patient and bed context

Representative engagement. Client details and internal figures are withheld under NDA. Compliance depends on how a system is operated as well as how it is built, and your own counsel decides what applies to you.