Hidden Markov Models for sepsis detection in preterm infants

TL;DR

This paper investigates the application of traditional and neural network-based hidden Markov models for early sepsis detection in preterm infants, demonstrating their potential advantages over other machine learning methods.

Contribution

It introduces a discriminative training approach for neural network-based HMMs and compares their performance with classical HMMs and other classifiers.

Findings

HMMs outperform logistic regression, SVM, and extreme learning machine in sepsis prediction.

Neural network-based HMMs with discriminative training show improved accuracy.

Classical Gaussian mixture model HMMs also demonstrate strong predictive capability.

Abstract

We explore the use of traditional and contemporary hidden Markov models (HMMs) for sequential physiological data analysis and sepsis prediction in preterm infants. We investigate the use of classical Gaussian mixture model based HMM, and a recently proposed neural network based HMM. To improve the neural network based HMM, we propose a discriminative training approach. Experimental results show the potential of HMMs over logistic regression, support vector machine and extreme learning machine.