Continuous Identification of Sepsis-Associated Acute Heart Failure Patients: An Integrated LSTM-Based Algorithm

Cardiovascular dysfunction often accompanies sepsis and increases the incidence of acute heart failure (AHF), which poses significant threats to patient survival and prognosis. Research applying machine learning to investigate AHF in this context is limited. To address this gap, we introduce a continuous model designed to assess AHF risk in real time among sepsis patients. Our framework integrates a static prediction model with continuous evaluation, yielding notable enhancements in performance. With an area under the receiver operating characteristic curve of 85.9 and an area under the precision–recall curve of 60.9, our algorithm outperformed traditional temporal modeling techniques and conventional machine learning methods. This research provides innovative perspectives and methodologies for AHF risk evaluation and offers substantial clinical utility.