# Responsible AI for Sepsis Prediction: Bridging the Gap Between Machine Learning Performance and Clinical Trust

## TL;DR

This paper explores how machine learning can help predict sepsis outcomes in ICUs, emphasizing the need for models that are accurate, fair, and explainable to gain doctors' trust.

## Contribution

The study evaluates and compares various machine learning models for sepsis prediction while emphasizing responsible AI and model interpretability.

## Key findings

- XGBoost outperformed other models in predicting hospital mortality with an AUROC of 0.874.
- Model interpretability using SHAP confirmed the clinical relevance of the variables used.
- Ensemble models showed strong predictive power but require explainability to be trusted in clinical settings.

## Abstract

Background: Sepsis remains a leading cause of mortality in intensive care units (ICUs) worldwide. Machine learning models for clinical prediction must be accurate, fair, transparent, and reliable to ensure that physicians feel confident in their decision-making processes. Methods: We used the MIMIC-IV (version 3.1) database to evaluate several machine learning architectures, including Logistic Regression, XGBoost, LightGBM, LSTM (Long Short-Term Memory) networks and Transformer models. We predicted three main clinical targets—hospital mortality, length of stay, and septic shock onset—using artificial intelligence algorithms, with respect for responsible AI principles. Model interpretability was assessed using Shapley Additive Explanations (SHAP). Results: The XGBoost model demonstrated superior performance in prediction tasks, particularly for hospital mortality (AUROC 0.874), outperforming traditional LSTM networks, Transformers, and linear baselines. The importance analysis of the variables confirmed the clinical relevance of the model. Conclusions: While XGBoost and ensemble algorithms demonstrate superior predictive power for sepsis prognosis, their clinical adoption necessitates robust explainability mechanisms to gain trust among doctors.