For Work / Against Work
Debates on the centrality of work

"Artificial Intelligence in Psychiatry"

by Fakhoury, Marc (2019)

Abstract

Scientific findings over the past few decades have shaped our understanding of the underlying neurobiology associated with psychiatric illnesses. However, despite significant advances in research, there is widespread disappointment with the overall pace of progress in detecting and treating psychiatric disorders. Current approaches for the diagnosis of psychiatric disorders largely rely on physician-patient questionnaires that are most of the time inaccurate and ineffective in providing a reliable assessment of symptoms. These limitations can, however, be overcome by applying artificial intelligence (AI) to electronic medical database and health records. AI in psychiatry is a general term that implies the use of computerized techniques and algorithms for the diagnosis, prevention, and treatment of mental illnesses. Although the past few years have witnessed an increase in the use of AI in the medical practice, its role in psychiatry remains a complex and unanswered question. This chapter provides the current state of knowledge of AI's use in the diagnosis, prediction, and treatment of psychiatric disorders, and examines the challenges and limitations of this approach in the medical practise.

Key Passage

AI-based techniques have also been effectively used in the prediction of psychiatric symptoms including psychosis, which broadly includes the manifestations of thought disorders, behavioral disorganization, or catatonia. Using automated speech analysis in combination with machine learning, Bedi et al. [9] were able to accurately predict the development of psychosis in high-risk youths, outperforming classifcation from clinical interviews, where much of the assessments rely on the patient’s motivation to accurately report his experience. Enhancing the capacity to predict psychosis could have signifcant impacts for the identifcation of high-risk individuals and could provide clinicians with valuable information on which to base treatment and prognostic decisions. Another application of machine learning algorithms is the prediction of suicide in high-risk individuals. Accounting to roughly 800 thousand deaths worldwide every year, suicide is a major public issue that cannot be ignored. Over the past few years, developments in machine learning techniques have proved effcient in determining with relatively high success the intent of suicide in high-risk individuals. For instance, machine learning algorithms based on linguistic and acoustic characteristics were successfully used to classify a cohort of subjects recruited from medical centers into suicidal, mentally ill but not suicidal, or control group with an accuracy of up to 85%. More recently, Walsh et al. [19] were able to accurately predict future suicide attempts in a cohort of adult patients with a history of self-injury by applying machine learning to electronic health records. Results have been more than 80% accurate in predicting whether someone will make a suicide attempt within the next two years, and 92% accurate in predicting whether someone will make a suicide attempt within the following 7 days. Last but not least, computerized text analytics applied to unstructured medical records predicted the risk of suicide in veterans with more than 65% accuracy, thereby allowing clinicians to better screen seemingly healthy individuals and to evaluate their risk for attempting suicide in the future. (p.122)

Keywords

Artificial Intelligence, Diagnosis, Language Processing, Machine Learning, Psychiatric Disorders

Themes

AI and Counselling, Automation

Links to Reference

Citation

Share


How to contribute.