Data-Driven Modeling and Analysis of Dynamic Cerebral Autoregulation

(collaborative project with Prof E. Miller and Prof R. Marshall, Division of Stroke and Cerebrovascular Disease, Dept. Neurology, Columbia University): 01/2018 - present

The project aims at developing data-driven techniques based on joint time-frequency analysis tools (e.g., [1]) and machine learning approaches for quantitative analysis of dynamic cerebral autoregulation, i.e., the ability of the cerebral vasculature to regulate cerebral blood flow in response to rapid changes in blood pressure. Loss of dynamic cerebral autoregulation can lead to strokes through either hyperperfusion causing blood-brain barrier compromise and brain hemorrhage, or hypoperfusion causing ischemic strokes.  

The ultimate research objective is to propose a biomarker for indicating healthy versus impaired autoregulation function, and for predicting eventually life-threatening events based on relevant patient data. Our long-term goal is to operationalize the above mathematical approaches to design, beta-test and validate a diagnostic instrument which can reliably measure dynamic cerebral autoregulation in real time at the bedside. This instrument will incorporate transcranial Doppler, a non-invasive monitoring device, and software capable of analyzing and processing multiple continuous signals in real time.    

[1] Miller E., Dos Santos K. R. M., Marshall R. S., Kougioumtzoglou I. A., 2020. Joint time-frequency analysis of dynamic cerebral autoregulation via generalized harmonic wavelets, Physiological Measurementvol. 41: 024002: 1-11.