Boamente : A Natural Language Processing-Based Digital Phenotyping Tool for Smart Monitoring of Suicidal Ideation

• Published in: Healthcare (Basel) Vol. 10; no. 4; p. 698
• Main Authors: Diniz, Evandro J S, Fontenele, José E, de Oliveira, Adonias C, Bastos, Victor H, Teixeira, Silmar, Rabêlo, Ricardo L, Calçada, Dario B, Dos Santos, Renato M, de Oliveira, Ana K, Teles, Ariel S
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 08.04.2022; MDPI
• Subjects: Algorithms; Artificial intelligence; COVID-19; Deep learning; eHealth; Keyboards; Medical personnel; Mental disorders; Mental health; mobile application; Natural language processing; Neural networks; Prevention; Psychologists; Smartphones; Social networks; Suicidal ideation; Suicides & suicide attempts; Telemedicine; deep learning; natural language processing; mental health; mobile application; suicide; artificial intelligence; eHealth
• ISSN: 2227-9032; 2227-9032
• DOI: 10.3390/healthcare10040698

People at risk of suicide tend to be isolated and cannot share their thoughts. For this reason, suicidal ideation monitoring becomes a hard task. Therefore, people at risk of suicide need to be monitored in a manner capable of identifying if and when they have a suicidal ideation, enabling professionals to perform timely interventions. This study aimed to develop the tool, a solution that collects textual data from users' smartphones and identifies the existence of suicidal ideation. The solution has a virtual keyboard mobile application that passively collects user texts and sends them to a web platform to be processed. The platform classifies texts using natural language processing and a deep learning model to recognize suicidal ideation, and the results are presented to mental health professionals in dashboards. Text classification for sentiment analysis was implemented with different machine/deep learning algorithms. A validation study was conducted to identify the model with the best performance results. The BERTimbau Large model performed better, reaching a recall of 0.953 (accuracy: 0.955; precision: 0.961; F-score: 0.954; AUC: 0.954). The proposed tool demonstrated an ability to identify suicidal ideation from user texts, which enabled it to be experimented with in studies with professionals and their patients.