Deep Forest Ensemble Model:a Novel Strategy for Complex Medical Image Data
10.11783/j.issn.1002-3674.2025.04.006
- VernacularTitle:深度森林联合模型:一种新的复杂医学影像数据的策略
- Author:
Yi ZHOU
1
;
Fang SHAO
;
Dongfang YOU
Author Information
1. 南京医科大学公共卫生学院生物统计学系 211166
- Publication Type:Journal Article
- Keywords:
Deep forest ensemble model;
Sobol-MDA;
Higher-order interactions;
Complex medical classification
- From:
Chinese Journal of Health Statistics
2025;42(4):510-515
- CountryChina
- Language:Chinese
-
Abstract:
Objective To compare the predictive performance of the deep forest ensemble model,deep forest,and random forest in medical classification.Methods This study proposed a deep forest ensemble model that integrated Sobol-MDA(Sobol-mean decrease accuracy)with the cascading structure of deep forest and feature extraction capabilities of random forest.The model was applied to both simulation and real medical data analysis.The simulation experiments covered scenarios such as unbalanced outcome variables,nonlinear relationships,noise variables,multicollinearity,and interactions.The real data analysis was conducted on parotid MRI data to compare the performance of the models in terms of area under curve(AUC)and other metrics.Results In both the simulation and real data analysis,the deep forest ensemble model demonstrated superiority,especially in complex interaction scenarios,where its predictive performance significantly outperformed deep forest and random forest models.Conclusion Deep forest ensemble model shows significant advantages in addressing complex medical classification tasks.Its predictive performance outperforms traditional models.
