Deep learning in mental health outcome research: a scoping review

Abstract

Mental illnesses, such as depression, are highly prevalent and have been shown to impact an individual's physical health. Recently, artificial intelligence (AI) methods have been introduced to assist mental health providers, including psychiatrists and psychologists, for decision-making based on patients' historical data (e.g., medical records, behavioral data, social media usage, etc.). Deep learning (DL), as one of the most recent generation of AI technologies, has demonstrated superior performance in many real-world applications ranging from computer vision to healthcare. The goal of this study is to review existing research on applications of DL algorithms in mental health outcome research. Specifically, we first briefly overview the state-of-the-art DL techniques. Then we review the literature relevant to DL applications in mental health outcomes. According to the application scenarios, we categorize these relevant articles into four groups: diagnosis and prognosis based on clinical data, analysis of genetics and genomics data for understanding mental health conditions, vocal and visual expression data analysis for disease detection, and estimation of risk of mental illness using social media data. Finally, we discuss challenges in using DL algorithms to improve our understanding of mental health conditions and suggest several promising directions for their applications in improving mental health diagnosis and treatment.

Applications of artificial intelligence technologies on mental health research during COVID-19

“…Artificial intelligence (AI) technologies offer a wide range of opportunities to leverage advancements in data sciences in analyzing health records, behavioral data, social media contents, and outcomes data on mental health (Mak et al, 2019;Su et al, 2020). A scoping review of machine learning methods identified 300 articles reporting the use of several AI technologies such as vector machines, neural networks, latent Dirichlet allocation, decision trees, and clustering (Shatte et al, 2019).…”

“…As more studies are emerging on the psychological impacts of COVID-19, scholars and practitioners should consider several potential strategies to leverage AI technologies wherever appropriate. First, electronic patient data can be assessed to identify patterns of mental health problems among hospitalized individuals with COVID-19, whereas community-based estimates can be linked-to predict how similar psychological issues are prevalent across populations (Shatte et al, 2019;Su et al, 2020). Such insights may facilitate the development and implementation of psychosocial interventions for people affected by COVID-19.…”

Machine learning for suicide risk prediction in children and adolescents with electronic health records

“…Hence, it is evident that clinical practitioners need more than just clinical assessment tools to identify patients at risk of suicide. Recent efforts to apply machine learning with electronic health record (EHR) data to predict suicide risk in adult populations [18][19][20][21][22][23] have not only confirmed the importance of prominent risk factors for suicidal behavior identified in prior research [24][25][26][27] but also identified other characteristics leading to improved accuracy in suicide prediction compared to previous efforts 19,28,29 . However, there is scarcity of advanced risk prediction models that can be applied in clinical practice to the general population of children, adolescents, and young adults.…”

Mining genetic and transcriptomic data using machine learning approaches in Parkinson’s disease

“…A deep learning model usually appears to be a “black box” model due to its high complexity. Though preliminary studies have reported a greater computational capacity and flexibility of deep learning in genetics and genomics 38 , 75 as well as health care 76 , it’s encountering a larger challenge in model interpretation. There have two potential strategies addressing this issue: to measure changes in model output while involving systematic modification of the input 77 ; or to engage third-party tools to determine the feature contributions 78 .…”