Trajectories Like Mine: Machine Learning for Healthcare, Summer 2017

The machine learning problem of “trajectories like mine” is to efficiently find patients with physiological waveforms similar to a reference waveform. Once a similarity set is found, it can be exploited for future or diagnostic extrapolations to the patient of reference without model-based learning. One ML approach for retrieving “trajectories like mine” is locality sensitive hashing. We are interested in scalable and practical implementations of LSH extensions for prediction problems in EEG, ECG or arterial blood pressure (ABP).


Contact: Erik Hemberg