Artificial intelligence-driven decision tree model for predicting quality of life determinants in women with endometriosis

Alexandre Vallée; Jean-Marc Ayoubi

doi:10.1002/ijgo.70450

Artificial intelligence-driven decision tree model for predicting quality of life determinants in women with endometriosis

Int J Gynaecol Obstet. 2025 Aug 8. doi: 10.1002/ijgo.70450. Online ahead of print.

Authors

Alexandre Vallée^{1

2}, Jean-Marc Ayoubi^{3

4}

Affiliations

¹ Department of Epidemiology and Public Health, Foch Hospital, Suresnes, France.
² Département Universitaire de Santé Publique, Prévention, Observation, Territoires (SPOT), Université de Versailles, Saint-Quentin-En-Yvelines (UVSQ), Versailles, France.
³ Department of Obstetrics, Gynecology and Reproductive Medicine, Foch Hospital, Suresnes, France.
⁴ Medical School, University of Versailles, Saint-Quentin-En-Yvelines (UVSQ), Versailles, France.

PMID: 40778483
DOI: 10.1002/ijgo.70450

Abstract

Objective: Endometriosis significantly impacts the quality of life (QoL) of affected women due to its complex symptomatology. This study aimed to develop a decision tree-based model to identify the key determinants influencing QoL in women with endometriosis, incorporating clinical, psychological, and sociodemographic factors.

Methods: A cross-sectional survey was conducted among 1586 women with endometriosis in France. QoL was assessed using the endometriosis health profile-5 (EHP-5) questionnaire, and key predictive variables included pain progression, dyspareunia, infertility, body mass index (BMI, calculated as weight in kilograms divided by the square of height in meters) digestive symptoms, and educational level. A decision tree model was trained (70% of data) and validated (30%) to classify women into good or poor QoL categories, with performance evaluated using area under the curve (AUC), sensitivity, and specificity metrics.

Results: The decision tree model achieved an AUC of 0.88 (training) and 0.75 (validation), with precision (0.94), sensitivity (75%), and specificity (84%) in the training phase. The model identified 25 patient subgroups, where progressive pain, dyspareunia, high BMI, infertility, and digestive symptoms were major predictors of poor QoL (100% probability). Conversely, women with no pain progression, no dyspareunia, low BMI, and high education had a 0% probability of poor QoL.

Conclusion: This study demonstrates the clinical utility of decision tree models in stratifying endometriosis patients based on QoL determinants. These findings support personalized treatment approaches, emphasizing pain management, fertility counseling, and lifestyle modifications. Future research should focus on AI integration into clinical decision tools for improved patient care.

Keywords: BMI; artificial intelligence; decision tree model; dyspareunia; endometriosis; infertility; machine learning; pain; personalized medicine; predictive model; quality of life; women's health.