Clear the fog of negative emotions: A new challenge for intervention towards drug users.
Affective computing
Convolutional neural networks
Drug users
Facial expression recognition
Machine learning
Negative emotions
Journal
Journal of affective disorders
ISSN: 1573-2517
Titre abrégé: J Affect Disord
Pays: Netherlands
ID NLM: 7906073
Informations de publication
Date de publication:
01 11 2021
01 11 2021
Historique:
received:
25
01
2021
revised:
26
06
2021
accepted:
11
07
2021
pubmed:
27
7
2021
medline:
29
10
2021
entrez:
26
7
2021
Statut:
ppublish
Résumé
The psychological and emotional problems of drug users are a focus of research. However, quick and effective emotion assessment tools were scarce. We aimed to use facial expression recognition to assess the emotional states of drug users. Our study was conducted in Chengdu City, Sichuan Province, China from January 1, 2020 to June 30, 2020. The 69 drug users who were undergoing community-based treatment were recruited. We developed an app to collect their images and videos, and trained the deep learning model to assess their emotional states. We also explored the correlation between emotional changes and treatment time, and investigated the impact factors associated with emotional changes. Based on the continuous 6-month follow-up study, the emotional distribution of drug users was still dominated by negative emotions during community treatment (72.85%). Nevertheless, with the increase of treatment time, 17.39% of drug users' emotions were changing better. Results also showed that compared with the females, males were less likely to have positive emotion change. In addition, the females were more willing to read reply messages from social workers. The relatively short observation period could be extended, and voice data should be considered for analysis in the future. Social workers should pay attention to emotional states of drug users, and provide effective and gender-specific psychological interventions for them. In addition, as a more powerful "medicine", it is essential to strengthen the accessibility of humanistic care and services to help drug users maintain a positive attitude.
Sections du résumé
BACKGROUND
The psychological and emotional problems of drug users are a focus of research. However, quick and effective emotion assessment tools were scarce. We aimed to use facial expression recognition to assess the emotional states of drug users.
METHODS
Our study was conducted in Chengdu City, Sichuan Province, China from January 1, 2020 to June 30, 2020. The 69 drug users who were undergoing community-based treatment were recruited. We developed an app to collect their images and videos, and trained the deep learning model to assess their emotional states. We also explored the correlation between emotional changes and treatment time, and investigated the impact factors associated with emotional changes.
RESULTS
Based on the continuous 6-month follow-up study, the emotional distribution of drug users was still dominated by negative emotions during community treatment (72.85%). Nevertheless, with the increase of treatment time, 17.39% of drug users' emotions were changing better. Results also showed that compared with the females, males were less likely to have positive emotion change. In addition, the females were more willing to read reply messages from social workers.
LIMITATIONS
The relatively short observation period could be extended, and voice data should be considered for analysis in the future.
CONCLUSIONS
Social workers should pay attention to emotional states of drug users, and provide effective and gender-specific psychological interventions for them. In addition, as a more powerful "medicine", it is essential to strengthen the accessibility of humanistic care and services to help drug users maintain a positive attitude.
Identifiants
pubmed: 34311330
pii: S0165-0327(21)00708-4
doi: 10.1016/j.jad.2021.07.029
pii:
doi:
Types de publication
Journal Article
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
305-313Informations de copyright
Copyright © 2021 Elsevier B.V. All rights reserved.