Osteoporotic fractures prediction in Chinese postmenopausal women: a machine learning-based multi-dimensional approach

1

Osteoporotic fractures significantly impact postmenopausal women, with 49 million affected in China, highlighting a major global health burden.
2

This study analyzed 1,717 postmenopausal women, identifying 32 clinical variables to improve fracture risk prediction beyond bone mineral density.
3

The Random Forest machine learning model achieved the highest performance (AUC = 0.872), surpassing other models like Extra Trees and XGBoost.
4

Key predictors for osteoporotic fractures included BMD, serum chloride, age, albumin-to-globulin ratio, and neutrophil percentage.
5

Integrating BMD with biochemical and clinical indicators enhances fracture risk prediction, aiding in clinical screening and risk stratification.

Frontiers In Endocrinology

by Wei Zhu, Yang Guo, Jiang Shuai, Longwang Tan, Chuang Liu, Yongjun Jia, Chi Zhang, Kok-Yong Chin
April 17, 2026

1