Natural Language Processing of Psychiatric Clinical Notes

Project Summary

Suicide is the second leading cause of death in youth and despite a growing body of research, mental health concerns in youth are still prevalent and under-treated. Understanding and identifying the risk factors associated with suicide in youth experiencing mental health concerns is paramount to early intervention. 45% of patients are admitted annually for suicidality at BC Children’s Hospital (BCCH).

At admission and discharge, patients receive extensive diagnostic assessments largely recorded as long narrative clinical notes. The reports are difficult to incorporate in large-scale analyses, as manual information extraction is laborious and prone to error.

Natural Language Processing (NLP) methodology is increasingly applied to extract meaningful information from free-text style clinical notes. It transforms the unstructured information embedded in the text into categorical and numerical fields amenable to data analysis.

There have only been limited applications of NLP to the mental health field. Most mental health clinical documentation is done through long narratives, which does not fit well into structured fields.

 

Our study aimed at investigating if NLP can be used to predict suicidality from clinical notes.

In particular, we explored the utility of sentiment analysis, a branch of NLP used most often to identify and quantify the sentiment, feelings, or opinions associated with a topic.

Project Findings

We developed a psychiatry-relevant lexicon and identified specific categories of words, such as thought content and thought process that had significantly different polarity between suicidal and non-suicidal cases.

In addition, we demonstrated that the individual words and their associated polarity can be used as features in classification models and carry informative content to differentiate between suicidal and non-suicidal cases.

 

In conclusion, our study reveals that there is much value in applying NLP to psychiatric clinical notes and suicidal prediction.

Publications

George A, Johnson D, Carenini G, Eslami A, Ng R, Portales-Casamar E. Applications of Aspect-based Sentiment Analysis on Psychiatric Clinical Notes to Study Suicide in Youth. AMIA Jt Summits Transl Sci Proc. 2021 May 17;2021:229-237.

Presentations

“Text Mining Psychiatric Clinical Notes,” BCCH Research Education Program Poster Day, July 2018

“Text Mining Psychiatric Clinical Notes,” BCCH Evidence to Innovation (E2i) 2018 Research Day, November 2018

“Natural Language Processing of Psychiatric Clinical Notes,” 2019 Information Technology & Communications in Health (ITCH) Conference, February 2019

“Text Mining Psychiatric Clinical Notes,” 2019 UBC Multidisciplinary Undergraduate Research Conference, March 2019

“Artificial Intelligence explorations at the Child and Adolescent Psychiatric Emergency (CAPE) Unit,” BCCH Artificial Intelligence in Healthcare Workshop, April 2019

“Artificial Intelligence and Machine Learning in Pediatric Mental Health,” 39th Annual Canadian Academy of Child and Adolescent Psychiatry Conference, September 2019

“Applications of Aspect-based Sentiment Analysis on Psychiatric Clinical Notes to Study Suicide in Youth,”; E2i Fall Research Forum, November 2020

“Applications of Aspect-based Sentiment Analysis on Psychiatric Clinical Notes to Study Suicide in Youth,” AMIA Informatics Summit, March 2021

This project is part of the Data Science and Health Informatics Cluster.

Team

Elodie Portales-Casamar, Lead ✉︎
Ali Eslami, Co-I
Raymond Ng, Co-I
Giuseppe Carenini, Co-I
Ali Mussavi Rizi, Co-I
Ahmed Abura’ed, postdoctoral fellow
Yuqian Zhuang, data scientist
Ariel Qi, patient partner
Alison Taylor, patient partner
Omar Bseiso, patient partner

Alumni:
Rebecca Lin
Esther Lin
Amy George
Cindy Ou Yang