PhD Project Predicting and understanding persistent critically ill patients
Research group Health Informatics & Data Science Pillar
Primary Supervisor Douglas Pires
Secondary Supervisor(s) Daniel Capurro
Discipline Computing and Information Systems
Contact email email@example.com
Keywords Machine Learning, Health Informatics, Process mining, Electronic Medical Records, Intensive Care Data, Readmission to ICU, Mortality Prediction
Our group focuses on leveraging advances in health informatics and machine learning to navigate through the wealth of patient data in electronic medical records to develop computational tools to better inform and support patient treatment and management.
While the majority of patients admitted to the intensive care unit (ICU) are expected to have a short ICU length of stay (LoS), a small proportion will instead progress into a syndrome called Persistent Critical Illness (PerCI). Patients who develop PerCI, have an ongoing reliance on intensive care therapies. That is mainly determined by their chronic co-morbidities, rather than the reason for which they were first admitted to the ICU.
While only a small proportion of patients remain in the ICU for a prolonged amount of time, the ones who do show worse outcomes. Such outcomes account for a disproportionate number of ICU bed-days and, as a consequence, resources. The impact of this group is expected to be of even greater significance in low- and middle-income countries, where fewer ICU beds are available and resources may be limited.
This project will aim to use intensive care databases to assess whether patient information can be used to predict persistent critical illness and how PerCI patient trajectories compare with those that stay for shorter periods of time.