Arvind RatheeDeveloper & curious humanBack to selected work

Healthcare · 2021–2022

Bringing laboratory results into a care workflow.

Integrated two external laboratories and helped connect diagnostic results, patient records, and generated reports.

  • React
  • Node.js
  • MongoDB
  • Redux Thunk
  • Puppeteer
  • AWS S3
ProjectMayaMD

Context

MayaMD PPHM supported healthcare workers conducting rural screening and helped connect patients to diagnostic laboratory services.

My role

End-to-end implementation of the laboratory integration, with senior review, alongside other frontend and backend feature work.

The challenge

Two laboratory vendors returned different data formats, and individual test results could arrive at different times. Reports needed to reflect the results available for the correct patient.

The approach

  • Normalized responses from the two labs into a consistent internal structure while retaining raw responses.
  • Implemented backend polling for results and supported partial as well as completed diagnostic reports.
  • Generated PDFs using EJS and Puppeteer, storing completed final reports in S3 and exposing them through the React application.

Outcome

The integration connected external laboratory data to patient records and made both partial and completed reports available to users.

Learning in production

After a month of MERN training, I joined MayaMD in December 2021 as an intern. A senior developer initially provided close guidance and continued reviewing my work as I took on more independent assignments.

The React application supported field workers and administrators in rural health screening. I worked on forms, tables, report screens, and Node.js APIs using MongoDB, Material UI, Redux, and Redux Thunk.

From sample to result

My first independently owned end-to-end feature integrated two diagnostic laboratories. I matched sample identifiers to patient records, mapped different vendor responses into a common format, and retained the original responses.

A backend scheduled process retrieved results as they became available. Tests could finish at different times, so the workflow supported partial results as well as completed orders.

Reports that reflected the latest results

I used EJS templates and Puppeteer to generate screening and diagnostic PDFs. A user could request a report with the results available so far; once all results were complete, the final report was stored in S3 for later access.

This project brought together the frontend, external APIs, scheduled processing, data modeling, and document generation—and taught me to own a feature while working within a senior review process.