Prediction of the NASH through penalized mixture of logistic regression models

Abstract

In this paper, a suitable and interpretable diagnosis statistical model is proposed to predict the Non-Alcoholic Steatohepatitis (NASH) from near infrared spectrometry data. In this disease, unknown patients profiles are expected to lead to different diagnosis. The model has then to take into account the heterogeneity of the data and the dimension of the spectrometric data. To this end, we propose to fit a mixture on the joint distribution of the diagnosis binary variable and the covariates selected in the spectra. Because of the high dimension of the data, a penalized maximum likelihood estimator is considered. In practice, a twofold penalty on both regression coefficients and covariance parameters is imposed. Automatic selection criteria such as the AIC and BIC are used to select the amount of shrinkage and the number of clusters. Performance of the overall procedure is evaluated through a simulation study and its application on the NASH data set is analysed. The model leads to higher prediction performance than competitive methods and provides highly interpretable results.