Personalised thrombo-embolic risk prediction after endometrial cancer surgery: an explainable AI approach using SHAP.
An AI model accurately predicts blood clot risk after gynecologic cancer surgery, offering doctors a practical tool for patient management.
A SHAP-explainable SVM model trained on 841 multi-center perioperative records achieves AUC 0.828 for predicting postoperative lower extremity DVT after endometrial cancer surgery, with robust external validation at AUC 0.819 and a deployable web interface for real-time clinical use. The SHAP analysis reveals a non-linear D-dimer risk threshold and validates a concise 4-variable model, advancing practical AI integration into gynecological oncology perioperative care.
What the study was
- Study design
- Multi-center retrospective ML model with external validation and web-based deployment
- Population
- Endometrial cancer surgical patients
- Sample size
- 841
- Category
- Diagnostics
- Maturity
- Validated
- Journal
- NPJ Digital Medicine
Why it surfaced
Validated multi-center ML model with web deployment for a common perioperative complication in gynecologic cancer — near-term clinical implementability is clear. NPJ Digital Medicine is a high-quality AI-in-medicine journal. Score 6 keeps this STANDARD despite the flag.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.