Manuscript Access Guarded
Full publication text and data are reserved for approved academic members. Sign in or register to unlock access.
Loading content...
Publication Manuscript Restricted — Sign In to Unlock
To protect academic manuscripts and peer-reviewed works, full publications require administrative approval. Sign in or register to request access.
Full publication text and data are reserved for approved academic members. Sign in or register to unlock access.
Early identification of pediatric sepsis in Intensive Care Units (ICUs) remains a significant clinical challenge due to the rapid progression of physiological deterioration. This paper introduces an optimized transformer-based neural architecture designed to analyze multi-modal clinical time-series data. By incorporating self-attention mechanisms across varying temporal scales, our approach models complex physiological correlations over extended windows.Validated on clinical datasets, the proposed architecture achieves a predictive AUROC of 0.94, outperforming traditional recurrent networks and clinical scoring tools. These results highlight the potential of deep learning sequence modeling to augment real-time ICU diagnostic alert systems.
Early identification of pediatric sepsis in Intensive Care Units (ICUs) remains a significant clinical challenge due to the rapid progression of physiological deterioration. This paper introduces an optimized transformer-based neural architecture designed to analyze multi-modal clinical time-series data. By incorporating self-attention mechanisms across varying temporal scales, our approach models complex physiological correlations over extended windows.Validated on clinical datasets, the proposed architecture achieves a predictive AUROC of 0.94, outperforming traditional recurrent networks and clinical scoring tools. These results highlight the potential of deep learning sequence modeling to augment real-time ICU diagnostic alert systems.