Phastar: Automating Medical Term Coding

Industry: Healthcare & Life Sciences
Solution: NLP-powered verbatim medical term standardisation
Impact: Boosted accuracy from 60% to 90%, saving analysts hours of manual work

Overview
Phastar is a leading global contract research organisation (CRO), providing statistical and data science services to biotech, pharma, and medical companies. One of their time-consuming challenges: manually coding free-text medical terms to standardised dictionaries.

The Challenge
Phastar analysts were spending significant time matching verbatim medical terms to their corresponding Low-Level Terms (LLTs) for regulatory datasets. They needed an automated solution that could maintain high accuracy in a complex domain.

The Ishango Approach
The Ishango team developed a deep learning NLP model based on Sentence-BERT (S-BERT), trained to understand the semantic relationships between raw terms and standardised entries. The goal was to streamline this previously manual process without compromising on precision.

The Result
The solution boosted term-matching accuracy from 60% to 90%, saving analysts substantial time and freeing them to focus on higher-value tasks.