Mild traumatic brain injury (mTBI) is challenging to diagnose due to its subtle and transient symptoms, making noninvasive diagnostic tools crucial for early detection. This study explores the use of a custom ResNet deep learning model combined with the Grad-CAM interpretability technique for mTBI detection via speech analysis. Speech data were transformed into Mel-spectrograms and fed into the model for binary classification between concussed and control individuals. The Grad-CAM method provided insights into which frequency regions of the Mel-spectrogram were most important for the model's predictions, with higher-frequency regions identified as significant for the model in detecting mTBI. Using Monte Carlo Cross-Validation (MCCV), we evaluated 50 different subject train-test split configurations to gain insights into the model's performance stability and variability. This analysis can assess the model's ability to learn consistent patterns within the dataset and suggest potential generalization tendencies. The variability observed in the performance metrics distribution underscores the importance of robust evaluation methods, particularly when working with small datasets. The combination of deep learning, robust evaluation technique and interpretability in this study contributes to the development of clinically viable, speech-based tools for mTBI detection, with potential applications in sports and healthcare settings.

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