Combining clinical notes with structured electronic health records enhances the prediction of mental health crises.

An automatic prediction of mental health crises can improve caseload prioritization and enable preventative interventions, improving patient outcomes and reducing costs. We combine structured electronic health records (EHRs) with clinical notes from 59,750 de-identified patients to predict the risk of mental health crisis relapse within the next 28 days. The results suggest that an ensemble machine learning model that relies on structured EHRs and clinical notes when available, and relying solely on structured data when the notes are unavailable, offers superior performance over models trained with either of the two data streams alone. Furthermore, the study provides key takeaways related to the required amount of clinical notes to add value in predictive analytics. This study sheds light on the untapped potential of clinical notes in the prediction of mental health crises and highlights the importance of choosing an appropriate machine learning method to combine structured and unstructured EHRs.

The Efficacy of "Foundations," a Digital Mental Health App to Improve Mental Well-being During COVID-19: Proof-of-Principle Randomized Controlled Trial.

BACKGROUND: Against a long-term trend of increasing demand, the COVID-19 pandemic has led to a global rise in common mental disorders. Now more than ever, there is an urgent need for scalable, evidence-based interventions to support mental well-being. OBJECTIVE: The aim of this proof-of-principle study was to evaluate the efficacy of a mobile-based app in adults with self-reported symptoms of anxiety and stress in a randomized control trial that took place during the first wave of the COVID-19 pandemic in the United Kingdom. METHODS: Adults with mild to severe anxiety and moderate to high levels of perceived stress were randomized to either the intervention or control arm. Participants in the intervention arm were given access to the Foundations app for the duration of the 4-week study. All participants were required to self-report a range of validated measures of mental well-being (10-item Connor-Davidson Resilience scale [CD-RISC-10], 7-item Generalized Anxiety Disorder scale [GAD-7], Office of National Statistics Four Subjective Well-being Questions [ONS-4], World Health Organization-5 Well-Being Index [WHO-5]) and sleep (Minimal Insomnia Scale [MISS]) at baseline and at weeks 2 and 4. The self-reported measures of perceived stress (10-item Perceived Stress Score [PSS-10]) were obtained weekly. RESULTS: A total of 136 participants completed the study and were included in the final analysis. The intervention group (n=62) showed significant improvements compared to the control group (n=74) on measures of anxiety, with a mean GAD-7 score change from baseline of -1.35 (SD 4.43) and -0.23 (SD 3.24), respectively (t134=1.71, P=.04); resilience, with a mean change in CD-RISC score of 1.79 (SD 4.08) and -0.31 (SD 3.16), respectively (t134=-3.37, P<.001); sleep, with a mean MISS score change of -1.16 (SD 2.67) and -0.26 (SD 2.29), respectively (t134=2.13, P=.01); and mental well-being, with a mean WHO-5 score change of 1.53 (SD 5.30) and -0.23 (SD 4.20), respectively (t134=-2.16, P=.02), within 2 weeks of using Foundations, with further improvements emerging at week 4. Perceived stress was also reduced within the intervention group, although the difference did not reach statistical significance relative to the control group, with a PSS score change from baseline to week 2 of -2.94 (SD 6.84) and -2.05 (SD 5.34), respectively (t134= 0.84, P=.20). CONCLUSIONS: This study provides a proof of principle that the digital mental health app Foundations can improve measures of mental well-being, anxiety, resilience, and sleep within 2 weeks of use, with greater effects after 4 weeks. Foundations therefore offers potential as a scalable, cost-effective, and accessible solution to enhance mental well-being, even during times of crisis such as the COVID-19 pandemic. TRIAL REGISTRATION: OSF Registries osf.io/f6djb; https://osf.io/vm3xq.

On the Information Bottleneck Problems: Models, Connections, Applications and Information Theoretic Views.

This tutorial paper focuses on the variants of the bottleneck problem taking an information theoretic perspective and discusses practical methods to solve it, as well as its connection to coding and learning aspects. The intimate connections of this setting to remote source-coding under logarithmic loss distortion measure, information combining, common reconstruction, the Wyner-Ahlswede-Korner problem, the efficiency of investment information, as well as, generalization, variational inference, representation learning, autoencoders, and others are highlighted. We discuss its extension to the distributed information bottleneck problem with emphasis on the Gaussian model and highlight the basic connections to the uplink Cloud Radio Access Networks (CRAN) with oblivious processing. For this model, the optimal trade-offs between relevance (i.e., information) and complexity (i.e., rates) in the discrete and vector Gaussian frameworks is determined. In the concluding outlook, some interesting problems are mentioned such as the characterization of the optimal inputs ("features") distributions under power limitations maximizing the "relevance" for the Gaussian information bottleneck, under "complexity" constraints.