Litcius/Paper detail

An AI-Based Framework for Real-Time Patient Monitoring and Intelligent Treatment Recommendation in Critical Care Units

T Tamilvizhi, R Thirumalini, M Takshinyaa, Orugunda Dedeepya

20258 citationsDOI

Abstract

Intensive Care Units (ICUs) require uninterrupted monitoring and timely intervention to improve critical care outcomes. Traditional alert systems often rely on static thresholds for vital signs, causing false alarms and delayed responses. This paper introduces a real-time AIpowered framework for disease prediction and treatment support in ICU settings. The system utilizes a Random Forest model trained on simulated ICU records, capable of recognizing over 30 critical conditions. Predictions are linked to ICD-10 codes and paired with standard medical treatments. A web-based dashboard using Streamlit offers dynamic risk visualization with live updates every five seconds. Remote access is securely enabled through Ngrok. Tested on 10,000 samples, the system achieved a 94.2% prediction accuracy. With its modular design and support for clinical standards, this solution is well-suited for both hospital ICUs and telemedicine integration.

Topics & Concepts

Remote patient monitoringDashboardTelemedicineModular designMedical emergencyMedicineIntensive careComputer scienceIntervention (counseling)Vital signsVisualizationTroubleshootingIntensive care unitContinuous monitoringPatient safetyCritically illeHealthRisk assessmentSepsis Diagnosis and TreatmentHealthcare Technology and Patient MonitoringMachine Learning in Healthcare