Patient monitoring involves repeated or continuous measurement of physiological parameters to provide clinicians with insights into a patient鈥檚 health status. This includes monitoring heart rate and rhythm, respiratory rate, blood pressure, and blood-oxygen saturation, among others. It plays a crucial role in critical care but also in home monitoring of at-risk patients. Despite advances in medical care, post-operative complications remain a concern, with up to 17% of surgical patients experiencing serious complications. Mechanical ventilation is often required in cases of respiratory failure, but challenges such as patient-ventilator asynchrony can lead to lung injury and increased mortality. Additionally, timely transitions between different levels of care are critical to optimize resource use and improve patient outcomes. AI-based predictive models have the potential to support these transitions by processing data from various sources at high frequencies to generate actionable insights. However, their successful integration into clinical workflows remains a challenge.
The BM/d lab develops statistical models for early prediction of patient deterioration and patient transition decisions, combining model-driven and data-driven approaches to improve clinical decision-making. Our research focuses on lung-protective ventilation strategies, including the detection and classification of patient-ventilator asynchronies to optimize mechanical ventilation. We investigate automated multi-parameter monitoring systems using wearable prototypes that collect ECG, heart rate, oxygen saturation, respiratory rate, and non-invasive blood pressure, alongside activity data to reduce false alarms and optimize workflows. In collaboration with clinical partners, we prioritize interpretable methods that incorporate domain knowledge while providing novel insights. Additionally, we explore non-invasive, non-obtrusive solutions for long-term vascular monitoring, focusing on technologies suitable for both hospital and home settings. This includes continuous hemodynamic monitoring via Doppler ultrasound, assessment of vascular stiffness, and modeling the physiological transfer function between peripheral and central circulation. Further, respiratory efforts are analyzed through electromyography to optimize assisted ventilation, and sweat metabolite analysis using advanced sensors is employed to infer blood concentrations.
Our patient monitoring research is conducted in collaboration with Catharina Hospital Eindhoven (the Netherlands) and Philips, ensuring close alignment with clinical needs and real-world implementation.