Maximum likelihood estimation for nonembeddable Markov chains when the cycle length is shorter than the data observation interval

by   Duncan Ermini Leaf, et al.
University of Southern California

Time-homogeneous Markov chains are often used as disease progression models in studies of cost-effectiveness and optimal decision-making. Maximum likelihood estimation of these models can be challenging when data are collected at a time interval longer than the model's transition cycle length. For example, it may be necessary to estimate a monthly transition model from data collected annually. The likelihood for a time-homogeneous Markov chain with transition matrix 𝐏 and data observed at intervals of T cycles is a function of 𝐏^T. The maximum likelihood estimate of 𝐏^T is easily obtained from the data. The Tth root of this estimate would then be a maximum likelihood estimate for 𝐏. However, the Tth root of 𝐏^T is not necessarily a valid transition matrix. Maximum likelihood estimation of 𝐏 is a constrained optimization problem when a valid root is unavailable. The optimization problem is not convex. Local convergence is explored in several case studies through graphical representations of a grid search. The example cases use disease progression data from the literature as well as synthetic data. The global maximum likelihood estimate is increasingly difficult to locate as the number of cycles or the number of states increases. What seems like a straightforward estimation problem can be challenging even for relatively simple models. Researchers should consider alternatives to likelihood maximization or alternative models.


page 13

page 15

page 17

page 20

page 34

page 36

page 38

page 39


Penalised maximum likelihood estimation in multistate models for interval-censored data

Multistate models can be used to describe transitions over time across s...

Unified Method for Markov Chain Transition Model Estimation Using Incomplete Survey Data

The Future Elderly Model and related microsimulations are modeled as Mar...

PCMC-Net: Feature-based Pairwise Choice Markov Chains

Pairwise Choice Markov Chains (PCMC) have been recently introduced to ov...

On a family of stochastic SVIR influenza epidemic models and maximum likelihood estimation

This study presents a family of stochastic models for the dynamics of in...

Learning Markov models via low-rank optimization

Modeling unknown systems from data is a precursor of system optimization...

Copula-based semiparametric transformation model for bivariate data under general interval censoring

This research is motivated by discovering and underpinning genetic cause...

Please sign up or login with your details

Forgot password? Click here to reset