LargeSolo build2023 — 2025

Hospital Management System

Registration to diagnosis to an SMS report link.

Full Stack EngineerSelf-hosted at hospital
Illustration representing Hospital Management System

The problem

A hospital running medical check-ups on paper: patients queued at every desk, results were typed twice, and reports were collected in person days later. The system had to follow the real physical flow of a patient through the building — not an idealised one.

A full clinical workflow — from patient registration to lab results and auto-generated reports delivered by SMS — used daily by a hospital.

What I built

  • Patient journey

    Registration, money receipt, sample collection and test assignment as one connected flow, each step visible to the next desk.

  • Pathology, radiology & physical tests

    Dedicated entry modules where doctors record observations against the exact sample and test that was ordered.

  • Automatic fit / unfit verdict

    Rules over the entered results decide the verdict and surface which parameters caused it, instead of leaving it to manual reading.

  • X-ray annotation

    Radiologists draw directly on the image to mark the problem area; the annotation is stored with the report.

  • Report delivery by SMS

    The final report is generated automatically and the patient receives an SMS with a link — no return trip to the hospital.

  • Biometric & passport capture

    Fingerprint identification for returning patients and a WebSocket-driven passport scanner that auto-fills registration forms.

Hard parts

  1. 01

    Hardware that lives outside the browser

    Passport scanners and biometric devices talk to a local machine, not a web page. I bridged them over WebSocket so the browser could receive scans instantly while keeping the desk workflow uninterrupted.

  2. 02

    Annotating medical images accurately

    Drawing on an X-ray has to survive zoom, different image sizes and re-opening the report. I stored annotations as coordinates relative to the image rather than baked-in pixels.

  3. 03

    Stability under daily load

    The platform served thousands of users a day. I refactored the diagnosis flow for correctness and stability, since a wrong or missing result is a clinical problem, not just a bug.

My contribution

  • Patient registration, money receipt, and sample collection flow, wired end-to-end.
  • Pathology, radiology, and physical test modules where doctors enter observations from collected samples.
  • Auto-generated fit / unfit verdicts and issue detection based on doctor-entered results.
  • X-ray annotation tool letting radiologists draw the problem area directly on the image.
  • Auto-generated patient-facing reports with an SMS delivering a secure report link to the patient.
  • Biometric patient identification and a WebSocket-based passport scanner for instant form auto-fill.

Outcome

  • Patients receive reports by SMS link instead of returning to collect them.
  • Verdicts and issue detection generated automatically from entered results.
  • Registration time cut by scanner and biometric auto-fill.
  • React
  • Express.js
  • MySQL
  • WebSocket