Machine learning for oral frailty factors in hospitalized schizophrenia patients: two-stage feature selection and SHAP analysis

1

The prevalence of oral frailty (OF) among long-term hospitalized patients with schizophrenia is 69.3%.
2

Machine learning models, particularly random forest, improved predictive accuracy for OF after two-stage feature selection.
3

Core risk factors for OF identified include Number of Teeth, Number of Psychiatric Hospitalizations, and Self-discontinuation of Medication.
4

Traditional regression models are insufficient for accurately assessing OF risk due to their tendency to overfit high-dimensional data.
5

The study emphasizes the need for individualized oral intervention programs based on identified determinants of OF in this population.

Frontiers In Psychiatry

by Yue Fu, Xing Yang, Tao Zhang, Yingying Wang, Shihan Tang, Zheng Luo, Cui Yang, Dongmei Wu
April 20, 2026

1