Source-linked AI summary
Explainable artificial intelligence model to predict acute critical illness from electronic health records
Simon Meyer Lauritsen, Mads Kristensen, Mathias Vassard Olsen, Morten Skaarup Larsen, Katrine Meyer Lauritsen, Marianne Johansson Jørgensen, Jeppe Lange, Bo Thiesson
TL;DR
Acute critical-illness prediction needs both strong performance and explanations clinicians can inspect. The xAI-EWS combines a temporal convolutional predictor with Deep Taylor Decomposition, achieving reported predictive performance while identifying relevant EHR inputs for sepsis, AKI, and ALI.
Problem
Existing AI systems can predict acute critical illness, but their complex decisions are difficult for clinicians to explain and assess.
Method
The xAI-EWS uses a temporal convolutional network to predict sepsis, AKI, and ALI from EHR data, with Deep Taylor Decomposition attributing predictions to input variables.
Results
AUROC values across onset to 24 hours before onset were 0.92–0.80 for sepsis, 0.88–0.79 for AKI, and 0.90–0.84 for ALI; explanations pinpointed relevant clinical parameters.
Takeaways & Limitations
The system provides individual and population-based explanations intended to help clinicians understand prediction reasoning and support transparency.
Takeaways & Limitations
The model’s predictive performance still requires validation in a different population, and better ground-truth definitions are needed for illnesses such as AKI and ALI.
Abstract
from arXiv · showhide
We developed an explainable artificial intelligence (AI) early warning score (xAI-EWS) system for early detection of acute critical illness. While maintaining a high predictive performance, our system explains to the clinician on which relevant electronic health records (EHRs) data the prediction is grounded. Acute critical illness is often preceded by deterioration of routinely measured clinical parameters, e.g., blood pressure and heart rate. Early clinical prediction is typically based on manually calculated screening metrics that simply weigh these parameters, such as Early Warning Scores (EWS). The predictive performance of EWSs yields a tradeoff between sensitivity and specificity that can lead to negative outcomes for the patient. Previous work on EHR-trained AI systems offers promising results with high levels of predictive performance in relation to the early, real-time prediction of acute critical illness. However, without insight into the complex decisions by such system, clinical translation is hindered. In this letter, we present our xAI-EWS system, which potentiates clinical translation by accompanying a prediction with information on the EHR data explaining it.