Dynamic AI and machine learning predictive models in intensive care: Transforming patient care in the ICU

Naoki Takegami, Department of Neurological Surgery, University of California San Francisco, San Francisco, USA
Shraddha Mainali, Department of Neurology, Virginia Commonwealth University, Richmond, USA

This article reviews the integration of Artificial Intelligence (AI) and Machine Learning (ML) in Intensive Care Units (ICU), with a particular focus on dynamic models. The ICU environment, rich in real-time data from monitors and electronic health records, is ideal for applying these models, which leverage the continuously evolving data in critical care to improve patient monitoring and decision-making. Since 2020, advancements in AI, data science, and computational power have enabled the development of dynamic models aimed at early detection, prediction, and monitoring of medical conditions, facilitating timely clinical interventions. The article also discusses the broader application of ML in healthcare, including diagnosis, prognosis, and treatment planning, acknowledging challenges related to data governance, ethics, and the need for interpretable and reliable models. It emphasizes the importance of a multidisciplinary approach that combines systems engineering, AI, and clinical expertise, with the goal of developing dynamic systems that enhance the quality and effectiveness of care in ICUs.

Keywords: Artificial Intelligence (AI). Machine Learning (ML). Intensive Care Units (ICU). Dynamic models. Early detection. Healthcare.