Supported Research

In partnership with Can-SOLVE

Co-Applicant(s):  Annie-Claire Nadeau-Fredette, Frederic Baroz, Neila Mezghani, Rita Suri, Ron Wald, Thomas Mavrakanas

Lay Abstract

Blood pressure disturbances during hemodialysis are frequent events (>10% of all sessions) that are associated with adverse health consequences including a higher risk of mortality, stroke, and dementia. These blood pressure variations are also the source of significant distress since they are accompanied by cramps, dizziness, loss of consciousness, anxiety, and fatigue that is prolonged after dialysis. It is important to prevent these events to improve both the physical and mental health of persons living with end-stage kidney disease. However, our preventive interventions are limited. This is in part because we have an incomplete understanding of how the body adapts to dialysis treatment to maintain adequate blood pressure. We propose a project that will use data automatically collected during hemodialysis sessions in 4 centers between 2017 and 2022 (1 million sessions in more than 1200 individuals) to deepen our understanding of how the body reacts during hemodialysis treatment using artificial intelligence methods. First, we will first use artificial intelligence (large language model) to identify sessions in which patients reported bothersome symptoms to their nurses during treatment and the type of symptoms they reported. This new method could be very helpful since information about symptoms during dialysis are rarely for research on large groups of patients since it is cumbersome for healthcare staff and patients to reliably collect. Secondly, we will then use another type of artificial intelligence (latent class analysis) that will identify patterns related to vital signs and symptoms during hemodialysis. Our hypothesis is that multiple patterns will be able to be identified and characterized for the first time. We will provide a detailed description of the characteristics of patients and sessions associated with these vital patterns to understand their meaning. Finally, we will then investigate what these patterns mean in terms of the risk of cardiovascular events, hospitalization, mortality and cognitive decline. The results of this project will provide important information to orient future efforts related to the prevention and treatment of blood pressure disturbances and associated distress. Since the efficacy of these treatments is likely to be different between patients, identifying the pattern indicating how the body reacts during hemodialysis could enable the identification of individuals that are more likely to benefit from a particular treatment. This project proposes to fully use the immense wealth of data recorded in the electronic medical record using modern analytical techniques in order to improve our understanding of a very frequent issue faced by patients receiving hemodialysis.