OP041 Topic: AS04–Emerging Sciences, Methodologies, Big Data and Technology: IMPALA PROJECT: REVOLUTIONIZING PEDIATRIC CRITICAL CARE IN LOW-RESOURCE SETTINGS WITH PREDICTIVE MONITORING

Abstract

Aims & Objectives: Continuous patient monitoring in critical care is pivotal for early intervention, but challenges persist in low-resource settings for which current systems are not designed. Integrating risk prediction and decision support can simplify interpretation and facilitate earlier interventions. These innovations could be especially relevant in settings with limited trained professionals and resources. The IMPALA project aims to develop a vital signs monitoring system with predictive algorithms that bridges this gap. Methods: Between 2019-2023, the EDCTP-IMPALA project conducted three studies in Malawi. First a proof-of-principle pilot study was done to refine a monitoring module applying novel sensors. Secondly an existing monitoring system was improved and supported by a novel tablet-based platform which was applied in 800 critically ill children to develop predictive algorithms. Thirdly an existing predictive algorithm was improved and integrated into the monitoring system and its performance assessed in a 200-patient pilot study. Results: The monitoring system enhanced early detection capabilities, supporting trained clinicians in identifying deterioration in the first pilot. Implementation of this technology in a related project resulted in a significant 59% (p=0.022) reduction in pediatric ward mortality. The predictive algorithms reached AUC-ROC values reaching up to 0.94. The second pilot study reported positive feedback from healthcare professionals. Conclusions: The IMPALA project showcases the feasibility of deploying a robust, predictive monitoring platform in low-resource settings, significantly reducing pediatric mortality. The integration of innovative technology and big data analytics represents a transformative step in enhancing pediatric critical care, promising improved outcomes and paving the way for future advancements in resource-constrained environments.Keywords: Low resource setting, predictive algorithm, Vital Signs Monitoring, Machine Learning, Early Warning Score

Ähnliche Arbeiten