Data-Driven Interactive Visualisation Platform for Paediatric Intensive Care Using the PIC Database

Recently, the government pressured the National Health Service (NHS) to improve its productivity by at least 4% while cutting costs by at least 1% for the upcoming year [3]. Additionally, such issues negatively affect the paediatric intensive care unit (PICU) as they must make quick and critical decisions with the increasing number of patients and limited available beds. Therefore, we have developed an interactive visual analytics tool to explore multimodal data from the Paediatric Intensive Care (PIC) database as there is no [1]. Collected from the Children's Hospital of Zhejiang University in 2010 and 2018, the PIC database includes over 12,000 admissions, capturing detailed information on demographics, vital signs, laboratory results, medications, and clinical notes. Our tool integrates open-source Python-based data processing and visualisation libraries (Pandas, Seaborn, and Matplotlib) into an interactive timeline interface, allowing clinicians to explore and correlate variables across time and patient cohorts and allowing it to be reusable on any structured medical data. This tool draws on a structured pipeline adapted from methodologies established for MIMIC-IV, tailored for the paediatric context [2]. It presents clinicians with relevant patient data on demand, enables the identification of clinical trends, and allows informed timely decision-making. By improving the utilisation, availability and readability of sophisticated data, the solution will support better outcomes in PICUs and contribute directly to NHS goals of care modernisation and addressing system pressures [3].