# BOBM: an adaptive deep learning framework for extended-window sepsis prediction with cross-institutional generalizability

## TL;DR

This paper introduces BOBM, a deep learning model that improves early sepsis prediction and works well across different hospitals and timeframes.

## Contribution

BOBM introduces a novel deep learning framework with bidirectional optimization and ensemble learning for sepsis prediction.

## Key findings

- BOBM achieved high AUC scores (0.86–0.95) in the MIMIC-IV dataset across different prediction timeframes.
- The model showed superior performance (AUC: 0.978–0.982) on two external datasets, demonstrating strong generalizability.
- BOBM extends the intervention window for sepsis detection up to 28 days before clinical suspicion.

## Abstract

Sepsis, a life-threatening and costly condition, necessitates early detection for effective management. However, diagnostic and therapeutic delays are common. Furthermore, sepsis complexity and dynamics challenge existing early warning systems, impeding timely intervention.

In this study, we introduce the Bayes-Optimized Boosting-Mamba Tab Model (BOBM), a novel deep learning-based algorithm that incorporates a bidirectional optimization learning mechanism within an ensemble framework. We designed dynamic agent models to enhance computational efficiency and adaptability across diverse clinical scenarios. Additionally, the model architecture was optimized for different temporal phases to enable a more precise estimation of the timing and progression of sepsis risk. To improve interpretability, we implemented SHapley Additive exPlanations value analysis, providing clinicians with actionable insights into potential patterns of sepsis progression.

The experimental evaluation demonstrated that our model attained the highest AUC (AUC: 0.86–0.95) in the MIMIC-IV dataset across different temporal windows prior to sepsis onset, confirming its adaptability and robust generalizability across various prediction timeframes. Furthermore, the model demonstrated superior performance (AUC: 0.978–0.982) when validated on two independent large-scale external datasets.

In contrast to traditional prediction approaches, our model extends the potential intervention timeframe significantly, while incorporating dynamic monitoring capabilities spanning the 7- to 28-day window before clinical suspicion, thereby establishing a foundation for regional sepsis surveillance systems.