| Abstrakt: |
This paper introduces a novel framework for precise prediction of suicide risk utilizing advanced machine learning techniques within health informatics. Our approach integrates multiple base classifiers, including Random Forest, Gradient Boosting, Support Vector Machine, k-nearest Neighbors, and Naive Bayes, alongside a neural network ensemble. Through meticulous data pre-processing and rigorous model training, we uncover crucial insights concealed within extensive demographic, historical, and socioeconomic indicators. Experimental results demonstrate the exceptional efficacy of our approach, with the Neural Network Ensemble achieving the highest accuracy of 91% in comparison to the base classifiers. Moreover, our ensemble approach outperforms individual base classifiers, as evidenced by the accuracy of Random Forest (85%), Gradient Boosting (86%), Support Vector Machine (82%), k-nearest Neighbors (78%), and Naive Bayes (75%). This research signifies a significant advancement in precision healthcare interventions, offering promising avenues for addressing pressing mental health challenges along with the suicidal trauma detection issue worldwide. |
Full text from SpringerLink