Heterogeneous Network Approach to Predict Individuals’ Mental Health

Depression and anxiety are critical public health issues affecting millions of people around the world. To identify individuals who vulnerable to depression anxiety, predictive models have been built that typically utilize data from one source. Unlike these traditional models, in this study, we leverage a rich heterogeneous dataset University Notre Dame’s NetHealth study collected individuals’ (student participants’) social interaction via smartphones, health-related behavioral wearables (Fitbit), trait surveys. integrate different types information, model as information network (HIN). Then, redefine problem predicting mental conditions (depression or anxiety) novel manner, applying our HIN popular paradigm recommender system (RS), which is used predict preference person would give an item (e.g., movie book). In case, items states. We evaluate four state-of-the-art RS approaches. Also, prediction another type—that node classification (NC) HIN, evaluating process features under logistic regression proof-of-concept classifier. find NC methods produce more accurate predictions than using same non-network fashion well random-approach. best considered approaches outperforms all This first smartphone, wearable sensor, survey manner use on conditions.