AI-Enhanced Intact Mass Data Communication: Expert-Level Comment and Automated Visualization

Published:

Organization: Bristol Myers Squibb
Date: May 2025

The Challenge:
Manual interpretation and commenting on LC-MS intact mass results is time-consuming and requires expert insight. No existing solution provided dynamic visualization of data across a 96-well plate.

My Solution:
I developed an automated tool with two core functionalities:

  • AI-Driven Comment Generation:
    • Integrates Protein Metrics report data with plate map metadata
    • Sends data to GPT-4.1 via Azure OpenAI API using tailored prompts
    • Produces expert-level, concise, and accessible comments for each sample
  • Interactive 96-Well Plate Heatmap Visualization:
    • Processes Protein Metrics report data to create an HTML-format 96-well plate heatmap
    • Color-coded species categories for easy differentiation

Impact:

  • Automated a highly specialized task
  • Improved clarity and consistency of reports
  • Facilitated faster, more informed decision-making

Technologies Used:
Python, Azure OpenAI API, Prompt Engineering, HTML