DualStream-MTCA: A Hybrid Deep Learning Model for the Simultaneous Early Detection of Sepsis and Heart Failure in Adult Intensive Care
- 1Computer Science and Computer Engineering Programme, School of Science and Engineering, University of Kurdistan Hewler, Erbil, Kurdistan Region, Iraq
Abstract
ObjectiveSepsis and heart failure produce overlapping early physiological signs in intensive care but call for opposite treatments, yet existing models predict one condition at a time. We built one model for both.
MethodsDualStream-MTCA combines two bidirectional long short-term memory encoders, for 48 hours of hourly vital signs and 7 days of daily laboratory results, with multi-head cross-attention, gradient-boosted-tree leaf embeddings and two task heads trained under uncertainty weighting. A refined configuration adds per-window aggregates, static patient features, self-supervised encoder pretraining and a five-seed ensemble with Monte Carlo dropout. Development used 53,229 adult stays from MIMIC-IV version 3.1; the frozen models were applied to 102,695 stays from the eICU Collaborative Research Database.
ResultsAt a prediction time 48 hours after admission, the refined ensemble reached an area under the receiver operating characteristic curve of 0.867 for sepsis and 0.899 for heart failure internally, against 0.832 and 0.857 for the baseline model. Isotonic regression brought expected calibration error below 0.015. For positives whose onset followed the prediction time by 4 to 12 hours, the area dropped to 0.66-0.70. Externally, the refined ensemble scored 0.750 for sepsis and 0.786 for heart failure, against 0.711 and 0.821 for the baseline model.
ConclusionA single multi-task model matched its single-task variants on both conditions, but discrimination fell under strictly anticipatory prediction and across sites, where static features carried little usable signal.
SignificanceTo our knowledge, this is the first model built specifically to predict both conditions jointly, with calibration, decision-curve and external evaluation.
Related
- Deep learning model for simultaneous vertebral level localization and fracture detection on DXA vertebral fracture assessment images
- Pulmonary thromboembolism with prolonged pulmonary infarction during intensive decongestion for severe heart failure : A case report
- Cross-kingdom antimicrobial resistance profiling: A unified multiplex PCR assay for simultaneous bacterial and fungal detection
- Discriminatory power of organ-specific Sequential Organ Failure Assessment (SOFA) scores: Predictors of maternal mortality in obstetric intensive care units (ICU)
- Sequential Organ Failure Assessment Score as a Predictor of Intensive Care Unit Mortality in Patients With Maxillofacial Trauma: A Prospective Observational Study.
BibTeX
@misc{khdir2026dualstreammtca,
title = {{DualStream-MTCA: A Hybrid Deep Learning Model for the Simultaneous Early Detection of Sepsis and Heart Failure in Adult Intensive Care}},
author = {S. A. Khdir and O. H. Ahmed},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.10.26362699},
url = {https://www.medrxiv.org/content/10.64898/2026.09.10.26362699v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - DualStream-MTCA: A Hybrid Deep Learning Model for the Simultaneous Early Detection of Sepsis and Heart Failure in Adult Intensive Care
AU - S. A. Khdir
AU - O. H. Ahmed
PY - 2026
DA - 2026/09/10
PB - medRxiv
DO - 10.64898/2026.09.10.26362699
UR - https://www.medrxiv.org/content/10.64898/2026.09.10.26362699v1
AB - ObjectiveSepsis and heart failure produce overlapping early physiological signs in intensive care but call for opposite treatments, yet existing models predict one condition at a time. We built one model for both. MethodsDualStream-MTCA combines two bidirectional long short-term memory encoders, for 48 hours of hourly vital signs and 7 days of daily laboratory results, with multi-head cross-attention, gradient-boosted-tree leaf embeddings and two task heads trained under uncertainty weighting. A refined configuration adds per-window aggregates, static patient features, self-supervised encoder pretraining and a five-seed ensemble with Monte Carlo dropout. Development used 53,229 adult stays from MIMIC-IV version 3.1; the frozen models were applied to 102,695 stays from the eICU Collaborative Research Database. ResultsAt a prediction time 48 hours after admission, the refined ensemble reached an area under the receiver operating characteristic curve of 0.867 for sepsis and 0.899 for heart failure internally, against 0.832 and 0.857 for the baseline model. Isotonic regression brought expected calibration error below 0.015. For positives whose onset followed the prediction time by 4 to 12 hours, the area dropped to 0.66-0.70. Externally, the refined ensemble scored 0.750 for sepsis and 0.786 for heart failure, against 0.711 and 0.821 for the baseline model. ConclusionA single multi-task model matched its single-task variants on both conditions, but discrimination fell under strictly anticipatory prediction and across sites, where static features carried little usable signal. SignificanceTo our knowledge, this is the first model built specifically to predict both conditions jointly, with calibration, decision-curve and external evaluation.
ER -