This article explores the use of probabilistic classification, namely finite mixture modelling, for identification of complex disease phenotypes, given cross-sectional data. In particular, if focuses on posterior probabilities of subgroup membership, a standard output of finite mixture modelling, and how the quantification of uncertainty in these probabilities can lead to more detailed analyses. Using a Bayesian approach, we describe two practical uses of this uncertainty: (i) as a means of describing a person's membership to a single or multiple latent subgroups and (ii) as a means of describing identified subgroups by patient-centred covariates not included in model estimation. These proposed uses are demonstrated on a case study in Parkinson's disease (PD), where latent subgroups are identified using multiple symptoms from the Unified Parkinson's Disease Rating Scale (UPDRS).
HoehnMMYahrMD. Parkinsonism: onset, progression and mortality. Neurology1967; 17: 427–442.
4.
ZetuskyWJankovicJPirozzoloF. The heterogeneity of Parkinson's disease: clinical and prognostic implications. Neurology1985; 35: 522–526.
5.
JankovicJ. Variable expression of Parkinson's disease: a base-line analysis of the DATATOP cohort. The Parkinson Study Group. Neurology1990; 40: 1529–1534.
ReijndersJEhrtULousbergRAarslandDLeentjensA. The association between motor subtypes and psychopathology in Parkinson's disease. Parkinsonism Relat Disord2009; 15: 379–382.
8.
GrahamJSagarH. A data-driven approach to the study of heterogeneity in idiopathic Parkinson's disease: identification of three distinct subtypes. Mov Disord1999; 14: 10–20.
9.
LewisSFoltynieTBlackwellARobbinsTOwenABarkerR. Heterogeneity of Parkinson's disease in the early clinical stages using a data-driven approach. J Neurol, Neurosurg Psychiatry2005; 76: 343–348.
10.
van RoodenSVisserMVerbaanDMarinusJvan HiltenJ. Motor patterns in Parkinson's disease: A data-driven approach. Movement Disord2009; 24: 1042–1047.
11.
EverittBSHandDJ. Finite mixture distributions, London: Chapman and Hall, 1989.
12.
MarinJMMengersenKLRobertCP. Bayesian modelling and inference on mixtures of distributions. Handbook Stat2005; 25: 459–507.
13.
MuthénBMuthénL. Integrating person-centered and variable-centered analyses: Growth mixture modelling with latent trajectory classes. Alcohol Clin Exp Res2000; 24: 882–891.
GoodmanLA. The analysis of systems of qualitative variables when some of the variables are unobservable. Part IA modified latent structure approach. Am J Sociol1974a; 79: 1179–1259.
17.
GoodmanLA. Exploratory latent structure analysis using both identifiable and unidentifiable models. Biometrika1974b; 61: 215–231.
18.
EspelandMAHandelmanSL. Using latent class models to characterize and assess relative error in discrete measurements. Biometrics1989; 45: 587–599.
19.
FormannAK. Linear logistic latent class analysis for polytomous data. J Am Stat Assoc1992; 87: 476–486.
WalshCD. Latent Class Analysis Identification of Syndromes in Alzheimer's Disease: A Bayesian Approach. Metodoloski Zvezki: Advances in Methodology and Statistics2006; 3: 147–162.
22.
SpycherBSilvermanMBrookeAMinderCKuehniC. Distinguishing phenotypes of childhood wheeze and cough using latent class analysis. Eur Respir J2008; 31: 974–974.
23.
GradosMAMathewsC.ATourette SyndromeAssociation. Latent class analysis of Gilles de la Tourette syndrome using comorbidities: clinical and genetic implications. Biol Psychiatry2008; 64: 219–219.
24.
LeaskSVermuntJDoneDCrowTBlowsMBoksM. Beyond symptom dimensions: Schizophrenia risk factors for patient groups derived by latent class analysis. Schizophr Res2009; 115: 346–350.
25.
NylundKLAsparouhovTMuthénBO. Deciding on the number of classes in latent class analysis and growth mixture modelling: A Monte Carlo simulation study. Struct Equation Model2007; 14: 535–569.
26.
Clogg CC. Latent class models: Recent developments and prospects for the future. In: Arminger G, Clogg C and Sobel M (eds) Handbook of statistical methods for the social and behavioral sciences. New York: Plenum, 1995.
27.
AndersonT. An introduction to multivariate statistical analysis, Chichester: Wiley, 1984.
28.
NaginDSTremblayRE. Analyzing developmental trajectories of distinct but related behaviors: A group-based method. Psychol Methods2001; 6: 18–34.
29.
Hagenaars JA. Loglinear models with latent variables. London: Sage, 1993.
30.
RubinDB. Multiple imputation for item nonresponse in surveys, New York: Wiley, 1987.
Bandeen-RocheKMigliorettiDLZegerSLRathouzPJ. Latent Variable Regression for Multiple Discrete Outcomes. J Am Stat Assoc1997; 92: 1375–1386.
34.
FormannAK. Linear logistic latent class analysis. Biometr J1982; 24: 171–190.
35.
DaytonCMMacreadyGB. Concomitant-variable latent-class models. J Am Stat Assoc1988; 83: 173–178.
36.
van der HeijdenPGMDessensJBockenholtUI. Estimating the concomitant-variable latent-class model with the EM algorithm. J Educ Behav Stat1996; 21: 215–229.
37.
RoederKLynchKGNaginDS. Modeling Uncertainty in Latent Class Membership: A Case Study in Criminology. J Am Stat Assoc1999; 94: 766–776.
38.
ReboussinBAIpEHWolfsonM. Locally dependent latent class models with covariates: an application to under-age drinking in the USA. J R Stat Soc Ser A2008; 171: 877–897.
Frühwirth-SchnatterS. Finite mixture and Markov switching models, New York: Springer, 2006.
41.
CeleuxG. Bayesian inference for mixtures: The label-switching problem. COMPSTAT1998; 98: 227–232.
42.
CeleuxGHurnMRobertCP. Computational and inferential difficulties with mixture posterior distributions. J Am Stat Assoc2000; 95: 957–970.
43.
StephensM. Dealing with label switching in mixture models. J R Stat Soc: Series B2000; 62: 795–809.
44.
AkaikeH. Factor analysis and AIC. Psychometrika1987; 52: 317–332.
45.
SchwartzG. Estimating the dimension of a model. Ann Stat1978; 6: 461–464.
46.
ScloveS. Application of model-selection criteria to some problems in multivariate analysis. Psychometrika1987; 52: 333–343.
47.
Tofighi D and Enders CK. Identifying the correct number of classes in growth mixture models. In: Hancock G and Samuelsen K (eds) Advances in latent variable mixture models. Greenwich: Information Age Publishing Inc., 2008.
48.
YangCC. Evaluating latent class analysis models in qualitative phenotype identification. Comput Stat Data Anal2006; 50: 1090–1104.
49.
JeffreysH. Theory of probability, Oxford: Clarendon Press, 1961.
50.
KassRRafteryA. Bayes Factors. J Am Stat Assoc1995; 90: 773–795.
51.
McLachlanGPeelD. Finite Mixture Models, New York: Wiley, 2000.
52.
ChibS. Marginal likelihood from the Gibbs Output. J Am Stat Assoc1995; 90: 1313–1321.
53.
CeleuxGSoromenhoG. An entropy criterion for assessing the number of clusters in a mixture model. J Classif1996; 13: 195–212.
54.
BanfieldJDRafteryAE. Model-based Gaussian and non Gaussian clustering. Biometrics1993; 49: 803–821.
55.
GarrettESZegerSL. Latent class model diagnosis. Biometrics2000; 56: 1055–1067.
56.
ZhangNL. Hierarchical Latent Class Models for Cluster Analysis. J Machine Learn Res2004; 5: 697–723.
57.
HagenaarsJ. Latent structure models with direct effects between indicators: Local dependence models. Sociol Methods Res1988; 16: 692–698.
58.
SutherlandGTHallidayGMSilburnPA. Do polymorphisms in the familial Parkinsonism genes contribute to risk for sporadic Parkinson's disease?Move Disord2009; 24: 833–838.
59.
KongALiuJWongW. Sequential imputation and Bayesian missing data problems. J Am Stat Assoc1994; 89: 278–288.
