[2301.11146] Two-step interpretable modeling of Intensive Care Acquired Infections

Obtain a PDF of the paper titled Two-step interpretable modeling of Intensive Care Acquired Infections, by Giacomo Lancia and three different authors

Obtain PDF
HTML (experimental)

Summary:We current a novel methodology for integrating excessive decision longitudinal knowledge with the dynamic prediction capabilities of survival fashions. The purpose is two-fold: to enhance the predictive energy whereas sustaining interpretability of the fashions. To transcend the black field paradigm of synthetic neural networks, we suggest a parsimonious and strong semi-parametric method (i.e., a landmarking competing dangers mannequin) that mixes routinely collected low-resolution knowledge with predictive options extracted from a convolutional neural community, that was skilled on excessive decision time-dependent data. We then use saliency maps to research and clarify the additional predictive energy of this mannequin. As an instance our methodology, we give attention to healthcare-associated infections in sufferers admitted to an intensive care unit.