A novel longitudinal clustering approach to psychopathology across diagnostic entities in the hospital-based PsyCourse study.
Bipolar disorder
Cluster analysis
Machine learning
Psychopathology
Schizophrenia
Journal
Schizophrenia research
ISSN: 1573-2509
Titre abrégé: Schizophr Res
Pays: Netherlands
ID NLM: 8804207
Informations de publication
Date de publication:
06 2022
06 2022
Historique:
received:
16
11
2020
revised:
23
01
2022
accepted:
02
05
2022
pubmed:
15
5
2022
medline:
22
6
2022
entrez:
14
5
2022
Statut:
ppublish
Résumé
Biological research and clinical management in psychiatry face two major impediments: the high degree of overlap in psychopathology between diagnoses and the inherent heterogeneity with regard to severity. Here, we aim to stratify cases into homogeneous transdiagnostic subgroups using psychometric information with the ultimate aim of identifying individuals with higher risk for severe illness. 397 participants of the PsyCourse study with schizophrenia- or bipolar-spectrum diagnoses were prospectively phenotyped over 18 months. Factor analysis of mixed data of different rating scales and subsequent longitudinal clustering were used to cluster disease trajectories. Five clusters of longitudinal trajectories were identified in the psychopathologic dimensions. Clusters differed significantly with regard to Global Assessment of Functioning, disease course, and-in some cases-diagnosis while there were no significant differences regarding sex, age at baseline or onset, duration of illness, or polygenic burden for schizophrenia. Longitudinal clustering may aid in identifying transdiagnostic homogeneous subgroups of individuals with severe psychiatric disease.
Identifiants
pubmed: 35567871
pii: S0920-9964(22)00163-3
doi: 10.1016/j.schres.2022.05.001
pii:
doi:
Types de publication
Journal Article
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
29-38Informations de copyright
Copyright © 2022. Published by Elsevier B.V.