Utilizing histogram to identify patients using pressors for acute hypotension
T. C. T. Ho, X Chen · 2009
In our approach, we propose to process the mean arterial pressure recordings like a signal but yet do analysis like an image in order to differentiate recordings from patients using pressors to treat their acute hypotension and from patients with no such conditions. Mean arterial pressure values recorded over time is numerical representations over time and hence signal processing techniques can be used to on such recordings for filtering purposes. Onset of hypotension in a pre-recorded reading, characterized by exhibiting 90 % low mean arterial pressures values anytime within a 30 minutes window, is similar to images exhibiting certain percentage of colours within its draw space. Hence, it can be pattern-matched similarly in image processing using histograms. By combining the 2 methods together, we are able to differentiate the 2 groups of patients for the first event of the Physionet/Computers in Cardiology Challenge 2009. 1.