Unsupervised Machine Learning to Identify Depressive Subtypes
1
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
Objectives: This study evaluated an unsupervised machine learning method, latent Dirichlet allocation (LDA), as a methodfor identifying subtypes of depression within symptom data. Methods: Data from 18,314 depressed patients were used to createLDA models. The outcomes included future emergency presentations, crisis events, and behavioral problems. One modelwas chosen for further analysis based upon its potential as a clinically meaningful construct. The associations between patientgroups created with the final LDA model and outcomes were tested. These steps were repeated with a commonly-usedlatent variable model to provide additional context to the LDA results. Results: Five subtypes were identified using the finalLDA model. Prior to the outcome analysis, the subtypes were labeled based upon the symptom distributions they produced:psychotic, severe, mild, agitated, and anergic-apathetic. The patient groups largely aligned with the outcome data. For example,the psychotic and severe subgroups were more likely to have emergency presentations (odds ratio [OR] = 1.29; 95% confidenceinterval [CI], 1.17–1.43 and OR = 1.16; 95% CI, 1.05–1.29, respectively), whereas these outcomes were less likely inthe mild subgroup (OR = 0.86; 95% CI, 0.78–0.94). We found that the LDA subtypes were characterized by clusters of uniquesymptoms. This contrasted with the latent variable model subtypes, which were largely stratified by severity. Conclusions:This study suggests that LDA can surface clinically meaningful, qualitative subtypes. Future work could be incorporated intostudies concerning the biological bases of depression, thereby contributing to the development of new psychiatric therapeutics.
Abstract
Objectives: This study evaluated an unsupervised machine learning method, latent Dirichlet allocation (LDA), as a methodfor identifying subtypes of depression within symptom data. Methods: Data from 18,314 depressed patients were used to createLDA models. The outcomes included future emergency presentations, crisis events, and behavioral problems. One modelwas chosen for further analysis based upon its potential as a clinically meaningful construct. The associations between patientgroups created with the final LDA model and outcomes were tested. These steps were repeated with a commonly-usedlatent variable model to provide additional context to the LDA results. Results: Five subtypes were identified using the finalLDA model. Prior to the outcome analysis, the subtypes were labeled based upon the symptom distributions they produced:psychotic, severe, mild, agitated, and anergic-apathetic. The patient groups largely aligned with the outcome data. For example,the psychotic and severe subgroups were more likely to have emergency presentations (odds ratio [OR] = 1.29; 95% confidenceinterval [CI], 1.17–1.43 and OR = 1.16; 95% CI, 1.05–1.29, respectively), whereas these outcomes were less likely inthe mild subgroup (OR = 0.86; 95% CI, 0.78–0.94). We found that the LDA subtypes were characterized by clusters of uniquesymptoms. This contrasted with the latent variable model subtypes, which were largely stratified by severity. Conclusions:This study suggests that LDA can surface clinically meaningful, qualitative subtypes. Future work could be incorporated intostudies concerning the biological bases of depression, thereby contributing to the development of new psychiatric therapeutics.
참고문헌 (28)
- World Health Organization, "Depression and other common mental disorders: global health estimates", World Health Organization, ~ (2017)
- GBD 2017 Disease and Injury Incidence and Prevalence Collaborators, "Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990-2017: a systematic analysis for the global burden of disease study 2017", Lancet, 392(10159), pp.1789~1858 (2018)[DOI]
- Warden D, "The STAR*D Project results : a comprehensive review of findings", Curr Psychiatry Rep, 9(6), pp.449~459 (2007)[DOI]
- Kern DM, "Treatment patterns and sequences of pharmacotherapy for patients diagnosed with depression in the United States : 2014 through 2019", BMC Psychiatry, 20(1), pp.4~ (2020)[DOI]
- Insel T, "Research domain criteria(RDoC) : toward a new classification framework for research on mental disorders", Am J Psychiatry, 167(7), pp.748~751 (2010)[DOI]
- Hasler G, "Discovering endophenotypes for major depression", Neuropsychopharmacology, 29(10), pp.1765~1781 (2004)[DOI]
- Rush AJ, "The varied clinical presentations of major depressive disorder", J Clin Psychiatry, 68(Suppl 8), pp.4~10 (2007)
- van Loo HM, "Data-driven subtypes of major depressive disorder : a systematic review", BMC Med, 10, pp.156~ (2012)[DOI]
- Ulbricht CM, "The use of latent class analysis for identifying subtypes of depression : a systematic review", Psychiatry Res, 266, pp.228~246 (2018)[DOI]
- Marquand AF, "Beyond lumping and splitting : a review of computational approaches for stratifying psychiatric disorders", Biol Psychiatry Cogn Neurosci Neuroimaging, 1(5), pp.433~447 (2016)[DOI]
- Fernandes BS, "The new field of 'precision psychiatry'", BMC Med, 15(1), pp.80~ (2017)[DOI]
- Horwitz T, "A decade in psychiatric GWAS research", Mol Psychiatry, 24(3), pp.378~389 (2019)[DOI]
- Fried EI, "Depression is not a consistent syndrome : An investigation of unique symptom patterns in the STAR*D study", J Affect Disord, 172, pp.96~102 (2015)[DOI]
- Blei DM, "Latent Dirichlet allocation", J Mach Learn Res, 3, pp.993~1022 (2003)
- Mori M, "Using latent class analysis to identify hidden clinical phenotypes", JAMA, 324(7), pp.700~701 (2020)[DOI]
- Lamers F, "Identifying depressive subtypes in a large cohort study : results from the Netherlands Study of Depression and Anxiety(NESDA)", J Clin Psychiatry, 71(12), pp.1582~1589 (2010)[DOI]
- Sullivan PF, "Latent class analysis of lifetime depressive symptoms in the national comorbidity survey", Am J Psychiatry, 155(10), pp.1398~1406 (1998)[DOI]
- Perera G, "Cohort profile of the South London and Maudsley NHS foundation trust Biomedical Research Centre(SLaM BRC)case register : current status and recent enhancement of an Electronic Mental Health Record-derived data resource", BMJ Open, 6(3), pp.e008721~ (2016)[DOI]
- Fernandes AC, "Development and evaluation of a de-identification procedure for a case register sourced from mental health electronic records", BMC Med Inform Decis Mak, 13, pp.71~ (2013)[DOI]
- "Library of production-ready applications" (2020)[URL]
- Delaffon V, "Use of Health of the Nation Outcome Scales in psychiatry", Adv Psychiatr Treat, 18(3), pp.173~179 (2012)[DOI]
- Linzer DA, "poLCA : an R package for polytomous variable latent class analysis", J Stat Softw, 42(10), pp.1~29 (2011)[DOI]
- Pedregosa F, "Scikit-learn : machine learning in Python", J Mach Learn Res, 12, pp.2825~2830 (2011)
- Tolentino JC, "DSM-5 criteria and depression severity : implications for clinical practice", Front Psychiatry, 9, pp.450~ (2018)[DOI]
- Lowe B, "Comparative validity of three screening questionnaires for DSM-IV depressive disorders and physicians' diagnoses", J Affect Disord, 78(2), pp.131~140 (2004)[DOI]
- van Loo HM, "Problems with latent class analysis to detect data-driven subtypes of depression", Mol Psychiatry, 23(3), pp.495~496 (2018)[DOI]
- Kroenke K, "The PHQ-9 : validity of a brief depression severity measure", J Gen Intern Med, 16(9), pp.606~613 (2001)[DOI]
- Liu CH, "The prevalence and predictors of mental health diagnoses and suicide among U.S. college students: implications for addressing disparities in service use", Depress Anxiety, 36(1), pp.8~17 (2019)[DOI]