End to End Autoencoder MLP Framework for Sepsis Prediction

2025-08-26

AI Summary

Early sepsis identification is hindered by irregular sampling, high missingness, and limitations of manual feature engineering in electronic health record (EHR) time-series data. To address these challenges, we propose an end-to-end deep learning framework that integrates an unsupervised autoencoder with a multilayer perceptron (MLP). The framework incorporates explicit missingness indicators, customized downsampling, and a non-overlapping dynamic sliding-window mechanism to enable automatic feature learning and real-time risk prediction from heterogeneous, noisy ICU time-series data—without handcrafted features. Evaluated on three independent ICU cohorts, our method achieves accuracies of 74.6%, 80.6%, and 93.5%, respectively, significantly outperforming conventional machine learning models. These results demonstrate strong cross-center robustness and clinical deployability.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Reasoning under Uncertainty: Sequential Decision Making

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Sepsis is a life threatening condition that requires timely detection in intensive care settings. Traditional machine learning approaches, including Naive Bayes, Support Vector Machine (SVM), Random Forest, and XGBoost, often rely on manual feature engineering and struggle with irregular, incomplete time-series data commonly present in electronic health records. We introduce an end-to-end deep learning framework integrating an unsupervised autoencoder for automatic feature extraction with a multilayer perceptron classifier for binary sepsis risk prediction. To enhance clinical applicability, we implement a customized down sampling strategy that extracts high information density segments during training and a non-overlapping dynamic sliding window mechanism for real-time inference. Preprocessed time series data are represented as fixed dimension vectors with explicit missingness indicators, mitigating bias and noise. We validate our approach on three ICU cohorts. Our end-to-end model achieves accuracies of 74.6 percent, 80.6 percent, and 93.5 percent, respectively, consistently outperforming traditional machine learning baselines. These results demonstrate the framework's superior robustness, generalizability, and clinical utility for early sepsis detection across heterogeneous ICU environments.

Problem

• Automating sepsis prediction from irregular ICU time-series data
• Overcoming manual feature engineering limitations in traditional models
• Enhancing clinical applicability with real-time inference mechanisms

Innovation

• End-to-end autoencoder MLP for sepsis prediction
• Custom downsampling and sliding window strategy
• Fixed-dimension vectors with missingness indicators