Identification of suicidal ideation in major depressive disorder: A machine learning approach with multimodal digital features
Identifikation suizidaler Gedanken bei Major Depression: ein Machine-Learning-Ansatz mit multimodalen digitalen Merkmalen
PLOS Digital Health · 27.08.2026 · 5 (8) · e0001615
DOI: 10.1371/journal.pdig.0001615 ↗Auf einen Blick
Bei 99 Patient:innen mit Major Depression wurden multimodale digitale Merkmale mit Fragebögen und klinisch bewerteter suizidaler Ideation verglichen. Das beste Modell erreichte eine AUC von 0,79.
Fragestellung
Können multimodale digitale Merkmale suizidale Ideation bei Patient:innen mit Major Depression mit Machine Learning identifizieren?
Methode
99 Patient:innen mit Major Depression; HADS und TAS-20 sowie 7 Tage EMA per Smartphone-App; klinische SI-Bewertung per HDRS-Interview; 10 ML-Modelle nach TRIPOD+AI.
Ergebnisse
38 Patient:innen hatten SI, 61 nicht. CatBoost zeigte die beste Leistung mit AUC 0,79 (p < .001), Accuracy 0,79 und F1-Score 0,73. KNN, ANN und Naive Bayes waren vergleichbar.
Einschränkungen
Laut Autor:innen sind größere und diversere Datensätze nötig, um Generalisierbarkeit und Genauigkeit zu verbessern.
Schlussfolgerung der Autor:innen
Multimodale digitale Merkmale waren mit SI bei MDD assoziiert. Die ML-Modelle zeigten nur eine mäßige Leistung bei der Identifikation von SI.
Kernaussagen
- Beste Modellleistung: CatBoost mit AUC 0,79
- 38 von 99 Patient:innen hatten suizidale Ideation
- Multimodale Merkmale waren mit SI assoziiert
Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.
Kurz gefasst
Fachartikel in der Zeitschrift PLOS Digital Health, online erschienen am 27.08.2026. Verfasst von Marco Shing Yan Man und 9 weiteren Autor:innen. Beteiligte Einrichtung: Faculty of Medicine, The Chinese University of Hong Kong, Hong Kong, China. Die Arbeit nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Health and Medical Research Fund, General Research Fund, Innovation and Technology Support Programme.
Abstract (Originalsprache)
Current detection models for suicidal ideation (SI) among depressed patients have primarily relied on clinical and biological features. This study aims at enhancing the real-world applicability of these models by using multimodal digital features. Patients self-administered the Hospital Anxiety and Depression Scale (HADS) and 20-item Toronto Alexithymia Scale (TAS-20). Their multimodal features (facial, acoustic and linguistic features) were obtained using ecological momentary assessment (EMA) with our self-developed smartphone application over 7 days. The Structured Interview Guide for the Hamilton Depression Rating Scale (HDRS) were conducted to identify clinically rated SI. Mood description recordings obtained from EMA were processed through natural language processing, Openface, and OpenSmile to obtain linguistic, facial and acoustics features, respectively. Multimodal features and questionnaires were then used for training and testing machine learning algorithms. The study was conducted and reported under the TRIPOD+AI guideline. Among the recruited 99 patients with major depressive disorder (MDD), 38 (mean age = 49.3 ± 9.7 y, 76% female) had SI while 61 (mean age = 51.5 ± 11.5 y, 76% female) did not have SI. Among the 10 machine learning models evaluated, the CatBoost classifier demonstrated the strongest detection performance, achieving an AUC of 0.79 (p < .001), accuracy of 0.79 and F1-score of 0.73. K-Nearest Neighbours (KNN), Artificial Neural Network (ANN) and Naive Bayes models also provided comparable results. Multimodal features were correlated with SI among MDD patients. Machine learning algorithms have modest performance in identifying SI among patients with MDD using multimodal digital features, including facial expressions, vocal characteristics and language use. However, larger and more diverse datasets are needed to enhance the generalizability and accuracy of these machine learning approaches in real-world clinical settings.
Autor:innen und Institutionen
- Faculty of Medicine, The Chinese University of Hong Kong, Hong Kong, China.
- Li Chiu Kong Family Sleep Assessment Unit, Department of Psychiatry, The Chinese University of Hong Kong, Hong Kong, China.
- Li Ka Shing Institute of Health Sciences, Faculty of Medicine, The Chinese University of Hong Kong, Hong Kong, China.
- Department of Psychiatry, Fujian Medical University Affiliated Fuzhou Neuropsychiatric Hospital, Fuzhou, China.
- Gerald Choa Neuroscience Institute, The Chinese University of Hong Kong, Hong Kong, China.
ORCID-Profile: Marco Shing Yan Man, Kit Ying Chan, Joey Wing Yan Chan, Tim Man Ho Li
Förderung
- Health and Medical Research Fund
- General Research Fund
- Innovation and Technology Support Programme
- CUHK Direct Grant for Research
- CUHK IdeaBooster Fund
- CUHK Improvement on Competitiveness in Hiring New Faculties Funding Scheme
Zitieren
APA
Man, M. S. Y., Au, A. L. M., Cheng, C. C. W., Siew, M. K. Y., Liu, Y., Chan, K. Y., Chen, J., Chan, J. W. Y., Wing, Y. K., & Li, T. M. H. (2026). Identification of suicidal ideation in major depressive disorder: A machine learning approach with multimodal digital features. PLOS Digital Health, 5(8), e0001615. https://doi.org/10.1371/journal.pdig.0001615
Vancouver
Man MSY, Au ALM, Cheng CCW, Siew MKY, Liu Y, Chan KY, et al. Identification of suicidal ideation in major depressive disorder: A machine learning approach with multimodal digital features. PLOS Digital Health. 2026;5(8):e0001615. doi:10.1371/journal.pdig.0001615
BibTeX
@article{Man2026,
author = {Man, Marco Shing Yan and Au, Andy Lok Man and Cheng, Christopher Chi Wai and Siew, Mavis Ka Yan and Liu, Yuyun and Chan, Kit Ying and Chen, Jie and Chan, Joey Wing Yan and Wing, Yun Kwok and Li, Tim Man Ho},
title = {Identification of suicidal ideation in major depressive disorder: A machine learning approach with multimodal digital features},
journal = {PLOS Digital Health},
year = {2026},
volume = {5},
number = {8},
pages = {e0001615},
doi = {10.1371/journal.pdig.0001615}
}
EUVIMED