We use machine learning, mathematical modeling and experimental data analysis to explore how the hematologic and cardiovascular systems respond to inflammatory pressures (ischemia, infection, trauma, etc.).
ML techniques are used to identify patterns in high-throughput routine clinical data streams in critical care settings such as following open-heart surgery or heart attacks. We then use mathematical modelling and analysis to capture and explore the dynamic mechanistic responses of patients to these clinical perturbations. Combining these two approaches allows for novel insights which can be translated into clinical diagnostic and prognostic tools.
Below are some key publications in this theme (click image to access paper). A full list of publications can be found here.
Some Upcoming Studies
Analysis of blood production dynamics in patients with SARS-CoV-2 infection.
Identification of inflammatory phenotypes during recovery from acute inflammatory events.
Agent-based modelling of red blood cell dynamics to predict emerging hematologic disease.
Association of red blood cell distribution width with mortality risk in hospitalized adults with SARS-CoV-2 infection.
JAMA Network Open. 2020.
Data-driven physiologic thresholds for iron deficiency associated with hematologic decline.
American Journal of Hematology. 2020
Diminished reactive hematopoiesis and cardiac inflammation in a mouse model of recurrent myocardial infarction.
Journal of the American College of Cardiology. 2020
Boston Globe: https://www.bostonglobe.com/2020/09/23/nation/mgh-study-says-routine-blood-test-may-predict-covid-19-hospital-death-risk/
Web MD: https://www.webmd.com/lung/news/20200923/blood-test-could-spot-those-at-highest-risk-for-severe-covid-19?#1