Machine-learning model for predicting depression in second-hand smokers in cross-sectional data using the Korea National Health and Nutrition Examination Survey.

Objective: Depression among non-smokers at risk of second-hand smoke (SHS) exposure has been a neglected public health concern despite their vulnerability. The objective of this study was to develop high-performance machine-learning (ML) models for the prediction of depression in non-smokers and to identify important predictors of depression for second-hand smokers. Methods: ML algorithms were created using demographic and clinical data from the Korea National Health and Nutrition Examination Survey (KNHANES) participants from 2014, 2016, and 2018 (N = 11,463). The Patient Health Questionnaire was used to diagnose depression with a total score of 10 or higher. The final model was selected according to the area under the curve (AUC) or sensitivity. Shapley additive explanations (SHAP) were used to identify influential features. Results: The light gradient boosting machine (LGBM) with the highest positive predictive value (PPV; 0.646) was selected as the best model among the ML algorithms, whereas the support vector machine (SVM) had the highest AUC (0.900). The most influential factors identified using the LGBM were stress perception, followed by subjective health status and quality of life. Among the smoking-related features, urine cotinine levels were the most important, and no linear relationship existed between the smoking-related features and the values of SHAP. Conclusions: Compared with the previously developed ML models, our LGBM models achieved excellent and even superior performance in predicting depression among non-smokers at risk of SHS exposure, suggesting potential goals for depression-preventive interventions for non-smokers during public health crises.

Requirements for Trustworthy Artificial Intelligence and its Application in Healthcare.

OBJECTIVES: Artificial intelligence (AI) technologies are developing very rapidly in the medical field, but have yet to be actively used in actual clinical settings. Ensuring reliability is essential to disseminating technologies, necessitating a wide range of research and subsequent social consensus on requirements for trustworthy AI. METHODS: This review divided the requirements for trustworthy medical AI into explainability, fairness, privacy protection, and robustness, investigated research trends in the literature on AI in healthcare, and explored the criteria for trustworthy AI in the medical field. RESULTS: Explainability provides a basis for determining whether healthcare providers would refer to the output of an AI model, which requires the further development of explainable AI technology, evaluation methods, and user interfaces. For AI fairness, the primary task is to identify evaluation metrics optimized for the medical field. As for privacy and robustness, further development of technologies is needed, especially in defending training data or AI algorithms against adversarial attacks. CONCLUSIONS: In the future, detailed standards need to be established according to the issues that medical AI would solve or the clinical field where medical AI would be used. Furthermore, these criteria should be reflected in AI-related regulations, such as AI development guidelines and approval processes for medical devices.