“
TL;DR
Five machine learning algorithms—logistic regression, Auto-GLM, Auto-Random Forests, Auto-Deep Learning, and Auto-GBM—were trained after SMOTE balancing of the imbalanced outcome classes.
Informatics in Medicine Unlocked · 8 authors, 7 centres
This summary was generated by AI from a single paper. It has not been reviewed by a clinician and is not clinical advice. Verify against the source before acting on it.
Five machine learning algorithms—logistic regression, Auto-GLM, Auto-Random Forests, Auto-Deep Learning, and Auto-GBM—were trained after SMOTE balancing of the imbalanced outcome classes.
Five machine learning algorithms—logistic regression, Auto-GLM, Auto-Random Forests, Auto-Deep Learning, and Auto-GBM—were trained after SMOTE balancing of the imbalanced outcome classes. Auto-GBM achieved the best validation performance with accuracy 89.75% and AUC 95.57%. Across all models, 47 variables were identified as key predictors spanning sociodemographic characteristics, health status, and personal behavior habits. Health-related predictors included not being confident about fitness level, being deaf, or having been told one had cancer. The study identified cancer-related information trust, contact with medical staff, and cancer screening awareness as novel predictor domains. Limitations include potential multicollinearity among selected variables, lack of causal mechanism analysis, and reduced interpretability of ensemble and deep learning models.