← Back to Projects

Lab Analytics & Reporting Platform

Centralized Azure-based data platform unifying lab operations data across merged systems into live PowerBI dashboards used by clinical and executive leadership.

August 1, 2024
Azure Data Factory Databricks PowerBI ETL SQL Python OpenAI
Lab Analytics & Reporting Platform

Overview

After the acquisition of Lifescan Health by Simple Laboratories, the combined organization was running two separate Laboratory Information Systems, three on-prem VM environments, and multiple disconnected reporting workflows built on ad hoc Excel files. Leadership had no single view of operations.

I designed and built the data platform that changed that — centralizing all lab data in Azure Databricks and surfacing it through PowerBI dashboards used daily by clinical directors and the executive team.

Architecture

LIS Systems (2x)  →  Azure Data Factory  →  Azure Databricks
Billing Platform  →  ADF Pipelines       →  Delta Lake Tables
Mirth Connect     →                      →  PowerBI Dashboards
External APIs     →                      →  Ad-hoc Queries

Data flows from source systems through parameterized ADF pipelines into a layered Delta Lake architecture (raw → cleaned → aggregated), then into semantic models consumed by PowerBI.

Dashboards Built

  • Order Volume — daily/weekly LIS order counts by test type, facility, and payer
  • Turnaround Time (TAT) — median and P95 TAT by test type, flagging SLA breaches in real time
  • High-Priority Order Tracker — live view of STAT and critical-value orders across facilities
  • Service Utilization — volume trends by client, enabling revenue forecasting and sales conversations
  • Performance Metrics — QC pass rates, instrument uptime, and phlebotomy productivity

Integrated OpenAI’s GPT API to build a reverse-search tool over historical HL7 messages. Lab staff can query in plain English (“find all CBC results for patient X from March”) instead of writing HL7 segment queries. PHI handling was designed to comply with HIPAA — no patient data is transmitted to OpenAI; the model operates on de-identified structural metadata only.

Results

  • Replaced Excel-based ad hoc reports used by every department
  • Executive team has same-day visibility into lab operations vs. weekly lag
  • ADF pipelines process data from two LIS systems without duplication or schema conflicts

Stack

Azure Data Factory · Azure Databricks · Delta Lake · PowerBI · Python · PySpark · OpenAI API · SQL · HIPAA-compliant architecture

← All Projects Back to Home