Sensing Psychosis: Using Artificial Intelligence & Machine Learning to Advance Psychiatric Assessment-View Presentation Slides

View Presentation Slides

Pdf Summary

The document discusses the use of artificial intelligence (AI) and machine learning (ML) to improve psychiatric assessments, with a focus on sensing psychosis. Dr. Justin Baker, an associate professor at Harvard Medical School, outlines his disclosure of previous consulting engagements and notes the financial relationships have been mitigated. He highlights the relevance of integrating technology into psychiatric evaluations to gather information about mental health conditions over time.<br /><br />The document proposes using latent construct models, sensor networks, and multimodal data to enable a more comprehensive understanding of mental health. This involves tracking various biological factors and behavioral patterns, such as sleep and activity, through wearables and remote monitoring.<br /><br />The document also covers the benefits of using automated tools for assessing psychiatric conditions. These tools can capture interactions and analyze speech and facial expressions in real-time, thereby reducing the need for costly, time-consuming, and inconsistent traditional interviews.<br /><br />The future of psychiatric assessment is described as one that incorporates comprehensive phenotyping and AI/ML-based predictive analytics to improve both research and clinical practices. The document mentions ongoing studies, such as a year-long monitoring of individuals with bipolar disorder or depression, emphasizing the importance of linking self-reports, clinician evaluations, and objective outcomes like sleep.<br /><br />Additionally, the document touches on ethical, legal, and social implications of using such advanced techniques in medicine and raises questions about whether AI will eventually replace psychiatrists.<br /><br />Concluding sections address tools and project extensions related to the Accelerating Medicines Partnership for Schizophrenia, introducing deep phenotyping tools like LOCHNESS for syncing data and DPDash for managing phenotyping data, suggesting a move towards more standardized computational approaches in psychiatry.

Keywords

artificial intelligence

machine learning

psychiatric assessments

sensing psychosis

latent construct models

wearables

real-time analysis

phenotyping

predictive analytics

ethical implications