Real-Time Prediction of Acute Kidney Injury in the Intensive Care Unit Using EDGE-AI Platform

Acute kidney injury (AKI) is a common early stage of renal degeneration in Intensive Care Unit (ICU) patients and is usually diagnosed by medical professionals after 48 h. We propose that a weight sensor be installed on the patient's urine bag in order to calculate hourly urine output and integrate patient data. The Acute Kidney Injury Network examined the previous 30-h urine output. A deep learning model (CNN-LSTM) predicts the AKI risk rate in 6 h. If the risk is higher than the threshold, use the hospital's official machine to alarm medical professionals. The model evaluation criteria Area Under Curve (AUC) was 0.97 (±0.02), precision = 0.96, recall = 0.96. To predict AKI in ICU 6 h earlier by using edge AI of urine output alarms medical professionals. It will improve unsustainable monitoring and result in immediate treatment and a 15% reduction in dialysis rates and a 20% reduction in mortality rates.