A range of chronic diseases have a significant influence on each other and share common risk factors. Comorbidity, which shows the existence of two or more diseases interacting or triggering each other, is an important measure for actuarial valuations. The main proposal of the study is to model parallel interacting processes describing two or more chronic diseases by a combination of hidden Markov theory and copula function. This study introduces a coupled hidden Markov model with the bivariate discrete copula function in the hidden process. To estimate the parameters of the model and deal with the numerical intractability of the log-likelihood, we use a variational expectation maximization algorithm. To perform the variational expectation maximization algorithm, a lower bound of the model’s log-likelihood is defined, and estimators of the parameters are computed in the M-part. A possible numerical underflow occurring in the computation of forward–backward probabilities is solved. The simulation study is conducted for two different levels of association to assess the performance of the proposed model, resulting in satisfactory findings. The proposed model was applied to hospital appointment data from a private hospital. The model defines the dependency structure of unobserved disease data and its dynamics. The application results demonstrate that the model is useful for investigating disease comorbidity when only population dynamics over time and no clinical data are available.
KoczwaraB. Cancer and Chronic Conditions. Singapore: Springer, 2016.
2.
FeinsteinAR. The pre-therapeutic classification of co-morbidity in chronic disease. J Chron Dis1970; 23: 455–468.
3.
SatarianoWA. Comorbidities and cancer. In: Hunter CP, Johnson KA and Muss HB (eds) Cancer in the Elderly. 1st ed. CRC Press, 2000, pp. 486–508.
4.
Guisado-ClaveroMRoso-LlorachALópezz-JimenezT, et al. Multimorbidity patterns in the elderly: a prospective cohort study with cluster analysis. BMC Geriatr2018; 18: 1–11.
5.
HuangCFLiuJCHuangHC, et al. Longitudinal transition trajectory of gouty arthritis and its comorbidities: a population-based study. Rheumatol Int2017; 37: 313–322.
6.
ViolánCFernández-BertolínSGuisado-ClaveroM, et al. Five-year trajectories of multimorbidity patterns in an elderly Mediterranean population using hidden Markov models. Sci Rep-UK2020; 10: 1–11.
7.
BringmannLFVissersNWichersM, et al. A network approach to psychopathology: new insights into clinical longitudinal data. PLoS ONE2013; 8: e60188.
8.
GroenRNRyanOWigmanJT, et al. Comorbidity between depression and anxiety: assessing the role of bridge mental states in dynamic psychological networks. BMC Med2020; 18: 1–17.
9.
QianZAlaaABellotA, et al. Learning dynamic and personalized comorbidity networks from event data using deep diffusion processes. In: 23 rd International Conference on Artificial Intelligence and Statistics, Online, 26-28 August 2020, pp. 3295–3305.
10.
FaruquiSHAAlaeddiniAJaramilloCA, et al. Mining patterns of comorbidity evolution in patients with multiple chronic conditions using unsupervised multi-level temporal Bayesian network. PLoS ONE2018; 13: e0199768.
11.
LappenschaarMHommersomALucasPJ, et al. Multilevel Bayesian networks for the analysis of hierarchical health care data. Artif Intell Med2013; 57: 171–183.
12.
LappenschaarMHommersomALucasPJ, et al. Multilevel temporal Bayesian networks can model longitudinal change in multimorbidity. J Clin Epidemiol2013; 66: 1405–1416.
13.
MaagBFeuerriegelSKrausM, et al. Modeling longitudinal dynamics of comorbidities. In: Proceedings of the Conference on Health, Inference, and Learning (ed Ghassemi M), Virtual USA, 8–10 April 2021, pp. 222–235. New York: Association for Computing Machinery.
14.
Leiva-MurilloJRodrguezABaca-GarcaE. Visualization and prediction of disease interactions with continuous-time hidden Markov models. In: NIPS 2011 Workshop on Personalized Medicine (ed Shawe-Taylor J and Zemel R), Granada, Spain, 16-17 December 2011.
15.
HuangZDongWWangF, et al. Medical inpatient journey modeling and clustering: a Bayesian hidden Markov model based approach. In: AMIA Annual Symposium Proceedings, San Francisco, USA, 14-18 November 2015, pp. 649–658, American Medical Informatics Association.
16.
PowellGVermaALuoY, et al. Modeling chronic obstructive pulmonary disease progression using continuous-time hidden Markov models. St Heal T2019; 264: 920–924.
17.
LeeCYoonJVan Der SchaarM. Dynamic-deephit: a deep learning approach for dynamic survival analysis with competing risks based on longitudinal data. IEEE T Bio-Med Eng2019; 67: 122–133.
18.
GeenensG. Copula modeling for discrete random vectors. Dependence Modeling2020; 8: 417–440.
19.
YuleGU. On the methods of measuring association between two attributes. J R Stat Soc1912; 75: 579–652.
20.
GoodmanLAKruskalWH. Measures of association for cross classifications. In: Goodman LA and Kruskal WH (eds) Measures of association for cross classifications. Springer, 1979, pp. 2–34.
21.
MostellerF. Association and estimation in contingency tables. J Am Stat Assoc1968; 63: 1–28.
22.
WangYPhamH. Modeling the dependent competing risks with multiple degradation processes and random shock using time-varying copulas. IEEE T Reliab2011; 61: 13–22.
23.
KaishevVKDimitrovaDSHabermanS. Modelling the joint distribution of competing risks survival times using copula functions. Insur Math and Econ2007; 41: 339–361.
24.
ChenYH. Semiparametric marginal regression analysis for dependent competing risks under an assumed copula. J Roy Stat Soc B Met2010; 72: 235–251.
25.
EscarelaGCarriereJF. Fitting competing risks with an assumed copula. Stat Methods Med Res2003; 12: 333–349.
26.
LoSMWilkeRA. A copula model for dependent competing risks. J Roy Stat Soc C-Appl2010; 59: 359–376.
27.
StöberJHongHGCzadoC, et al. Comorbidity of chronic diseases in the elderly: patterns identified by a copula design for mixed responses. Comput Stat Data An2015; 88: 28–39.
28.
DempsterAPLairdNMRubinDB. Maximum likelihood from incomplete data via the EM algorithm. J Roy Stat Soc B Met1977; 39: 1–22.
29.
SaulLKJaakkolaTJordanMI. Mean field theory for sigmoid belief networks. J Artif Intell Res1996; 4: 61–76.
30.
JaakkolaT. Tutorial on variational approximation methods. In Advanced mean field methods: theory and practice, 2000, pp. 129–159. MIT Press.
31.
BleiDMKucukelbirAMcAuliffeJD. Variational inference: a review for statisticians. J Am Stat Assoc2017; 112: 859–877.
32.
WainwrightMJJordanMI. Graphical models, exponential families, and variational inference. Found Trends Mach Learn2008; 1: 1–305.
33.
OrmerodJTWandMP. Explaining variational approximations. Am Stat2010; 64: 140–153.
34.
OflazZYozgatligilCSelcuk-KestelAS. Estimation of disease progression for ischemic heart disease using latent Markov with covariates. Stat Anal Data Min: ASA Data Sci J2022; 16: 16–28.
35.
ValderasJMStarfieldBSibbaldB, et al. Defining comorbidity: implications for understanding health and health services. Ann Fam Med2009; 7: 357–363.
36.
SiuTKChingWKFungSE, et al. On a multivariate Markov chain model for credit risk measurement. Quant Financ2005; 5: 543–556.
37.
PasanisiAFuSBousquetN. Estimating discrete Markov models from various incomplete data schemes. Comput Stat Data An2012; 56: 2609–2625.
FineSSingerYTishbyN. The hierarchical hidden Markov model: analysis and applications. Mach Learn1998; 32: 41–62.
40.
GhahramaniZJordanMI. Factorial hidden Markov models. In: Advances in Neural Information Processing Systems 8 (ed. Touretzky D, Mozer MC and Hasselmo M), 1995, pp. 472–478.
41.
KristjanssonTTFreyBJHuangTS. Event-coupled hidden Markov models. In: 2000 IEEE International Conference on Multimedia and Expo, New York, USA, 30 July–2 August 2000, volume 1, pp. 385-388, IEEE.
42.
BrandMOliverNPentlandA. Coupled hidden Markov models for complex action recognition. In: Proceedings of IEEE computer society conference on computer vision and pattern recognition, San Juan, USA, 17–19 June 1997, pp. 994–999, IEEE.
43.
KwonJMurphyK. Modeling freeway traffic with coupled HMMs. Report. USA: University of California, 2000.
44.
ZhongSGhoshJ. A new formulation of coupled hidden Markov models. Report. USA: Dept Elect Comput Eng, University of Austin, 2001.
45.
Lapuyade-LahorgueJXueJHRuanS. Segmenting multi-source images using hidden Markov fields with copula-based multivariate statistical distributions. IEEE T Image Process2017; 26: 3187–3195.
46.
HuX. A Copula-based Quantile Risk Measure Approach to Hedging under Regime Switching. Master Thesis, University of Waterloo, Canada, 2015.
47.
DerrodeSPieczynskiW. Unsupervised classification using hidden Markov chain with unknown noise copulas and margins. Signal Process2016; 128: 8–17.
48.
LagonaF. Copula-based segmentation of cylindrical time series. Stat Probabil Lett2019; 144: 16–22.
49.
SunFJiangY. A hidden resource in wireless channel capacity: Dependence control in action. arXiv preprint arXiv:180500812 2018.
50.
SklarM. Fonctions de repartition an dimensions et leurs marges. Publ inst statist univ Paris1959; 8: 229–231.
51.
NelsenRB. An introduction to copulas, volume 139 of. Lecture Notes in Statistics 1999.
52.
WangXLebarbierEAubertJ, et al. Variational inference for coupled hidden Markov models applied to the joint detection of copy number variations. Int J Biostat2019; 15: 20180023.
NielsenFSunK. Guaranteed bounds on the Kullback-Leibler divergence of univariate mixtures using piecewise log-sum-exp inequalities. arXiv preprint arXiv:1606.05850 2016.
56.
AndersonTWGoodmanLA. Statistical inference about Markov chains. Ann Math Stat1957; 28: 89–110.
57.
SpedicatoGA. Discrete Time Markov Chains with R. R J 2017.
58.
ZucchiniWMacDonaldILLangrockR. Hidden Markov models for time series: an introduction using R. Chapman and Hall/CRC, 2017.
59.
KubanekMBobulskiJAdrjanowiczL. Characteristics of the use of coupled hidden Markov models for audio-visual polish speech recognition. B Pol Acad Sci-Tech2012; 60: 307–316.
60.
GhoshSLiJCaoL, et al. Septic shock prediction for ICU patients via coupled HMM walking on sequential contrast patterns. J Biomed Inform2017; 66: 19–31.
61.
DarmanjianSKimSPNechybaMC, et al. Independently coupled HMM switching classifier for a bimodel brain–machine interface. In: 2006 16th IEEE Signal Processing Society Workshop on Machine Learning for Signal Processing, Maynooth, Ireland, 6–8 September 2006, pp. 379–384, Piscataway, NJ IEEE Service Center.
62.
AbdelazizAHCharafLAZeilerS, et al. On dynamic stream weight learning for coupled-hmm-based audio-visual speech recognition. In: 40th Annual German Congress on Acoustics, Olderburg, Germany, 2014, pp. 531–532.
63.
WuJXunPTangQ, et al. Circulating magnesium levels and incidence of coronary heart diseases, hypertension, and type 2 diabetes mellitus: a meta-analysis of prospective cohort studies. Nutr J2017; 16: 1–13.
64.
SchellevisFGvan der VeldenJvan de LisdonkE, et al. Comorbidity of chronic diseases in general practice. J Clin Epidemiol1993; 46: 469–473.
65.
BeghettiMTissotC. Pulmonary arterial hypertension in congenital heart diseases. Semin Respir Crit Care Med2009; 30: 421–428.
66.
VachiéryJLAdirYBarberàJA, et al. Pulmonary hypertension due to left heart diseases. J Am Coll Cardiol2013; 62: D100–D108.
67.
JinPGuWLaiY, et al. The circulating microRNA-206 level predicts the severity of pulmonary hypertension in patients with left heart diseases. Cell Physiol Biochem2017; 41: 2150–2160.
68.
OflazZ. Coupled Hidden Markov Model with Bivariate Discrete Copula to Study Comorbidity of Chronic Diseases. PhD Thesis (Unpublished), Middle East Technical University, Turkey, 2022.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
