Two-step interpretable modeling of Intensive Care Acquired Infections

TL;DR

This paper introduces a semi-parametric, interpretable model that combines low-resolution clinical data with high-resolution features extracted via neural networks to improve prediction of ICU-acquired infections.

Contribution

The paper presents a novel two-step modeling approach integrating neural network features with survival analysis for better interpretability and predictive accuracy in ICU infection prediction.

Findings

Enhanced predictive performance over traditional models

Effective use of neural network features for interpretability

Application to healthcare-associated infections in ICU patients

Abstract

We present a novel methodology for integrating high resolution longitudinal data with the dynamic prediction capabilities of survival models. The aim is two-fold: to improve the predictive power while maintaining interpretability of the models. To go beyond the black box paradigm of artificial neural networks, we propose a parsimonious and robust semi-parametric approach (i.e., a landmarking competing risks model) that combines routinely collected low-resolution data with predictive features extracted from a convolutional neural network, that was trained on high resolution time-dependent information. We then use saliency maps to analyze and explain the extra predictive power of this model. To illustrate our methodology, we focus on healthcare-associated infections in patients admitted to an intensive care unit.