Research Breakthrough: A team of researchers at The University of Texas Health Science Center developed a pioneering AI tool to better predict treatment response, a step toward earlier personalized treatment plans.
The researchers created a new artificial intelligence tool that uses images from diagnostic laparoscopy, minimally invasive camera-made images taken inside the abdomen, to predict how patients with high-grade serous ovarian cancer (HGSOC) will respond to standard treatment. They trained a deep-learning model that looks at pre-treatment images and classifies patients into two groups: those likely to have a short progression-free survival (PFS) (less than 8 months) and those with a longer PFS (over 12 months). This model leverages advanced techniques like contrastive pre-training to learn image features, and a location-aware transformer to make predictions based solely on visual data.
The model performed impressively when tested across the full dataset; it could reliably distinguish between patients at higher or lower risk of early recurrence just from their laparoscopy images. If refined and validated further, this tool could help doctors better personalize treatment plans right at diagnosis, making early, tailored decisions possible without relying on complex genetic tests or waiting for long-term outcomes. It’s a promising step toward making ovarian cancer care smarter and faster, using existing diagnostic procedures paired with powerful AI.
Read more:
A pioneering artificial intelligence tool to predict treatment outcomes in ovarian cancer via diagnostic laparoscopy, published in Nature on April 25, 2025.
Ovarian Cancer Research Alliance (OCRA) Support
- This study was supported by OCRA’s 2023 Collaborative Research Development Grant – Microsoft AI for Good Lab, awarded to Shayan Shams, PhD.
- OCRA Grantee Co-Authors include Shayan Shams, PhD, The University of Texas Health Science Center; and Anil Sood, MD, The University of Texas MD Anderson Cancer Center.
