Use of natural language processing to identify inflammatory breast cancer cases across a healthcare system

Ramez Kouzy ¹, Megumi Kai ^2,3, Huong T Le-Petross ⁴

⁰Department of Radiation Oncology, Division of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA.

Jnci Cancer Spectrum +

|

June 10, 2025

View abstract on PubMed

Summary

Artificial intelligence and natural language processing (NLP) can identify rare inflammatory breast cancer (IBC) cases missed by traditional methods. This AI platform improves access to specialized care for patients with inflammatory breast cancer.

Area Of Science

Oncology
Medical Informatics
Artificial Intelligence

Background

Early identification of inflammatory breast cancer (IBC) is crucial but challenging in large healthcare systems.
Limited access to specialized care hinders timely diagnosis and treatment for IBC patients.
Current surveillance methods may miss rare but aggressive cancer cases.

Purpose Of The Study

To develop and evaluate an AI-driven platform using NLP and electronic health records (EHRs) for systematic IBC case identification.
To improve the detection rate of inflammatory breast cancer cases within a multi-campus healthcare network.
To assess the performance of NLP and human review in identifying rare oncological conditions.

Main Methods

An AI platform integrating NLP with EHRs was developed to screen clinical notes for potential IBC cases.
A sequential review process involved initial NLP screening, followed by human-in-the-loop validation and multidisciplinary confirmation.
The platform analyzed over 8.6 million clinical notes across five healthcare campuses.

Main Results

The AI platform demonstrated 92.2% sensitivity in identifying confirmed IBC cases.
It successfully identified 57 IBC cases (22.4%) missed by traditional surveillance methods.
Human-in-the-loop review significantly improved the positive predictive value from 55.4% to 78.4%.

Conclusions

Lightweight NLP systems combined with targeted human review can effectively identify rare cancer cases within complex healthcare networks.
This approach enhances access to specialized care for patients with conditions like inflammatory breast cancer.
The study serves as a proof-of-concept for using AI in rare disease detection and improving patient pathways.