1) Explaining Feature Representation as learned by Neural Networks and Convolution Networks:
The success of neural and deep algorithms on image and text data has increased the interest of various healthcare organizations to use AI for supporting better patient care. Although the performance accuracies of these data-based, representation-learning models are quite high, their reputation as computational ‘black-box’ raises questions of trustworthiness when used for developing decision-support systems and performing predictive analysis in healthcare and related fields. Thus, there is a critical need to identify how these networks internally represent or encode the features of a given data type. Otherwise, they remain a blackbox, reducing their wide-spread use. This project involves mathematical modeling, designing and running experiments, and running simulations.
Sepsis is a life-threatening condition that occurs when the body's response to infection causes tissue damage, organ failure, or death (Singer et al., 2016). In the U.S., nearly 1.7 million people develop sepsis and 270,000 people die from sepsis each year; over one-third of people who die in U.S. hospitals have sepsis (CDC). Internationally, an estimated 30 million people develop sepsis and 6 million people die from sepsis each year; an estimated 4.2 million newborns and children are affected (WHO). Sepsis costs U.S. hospitals more than any other health condition at $24 billion (13% of U.S. healthcare expenses) a year, and a majority of these costs are for sepsis patients that were not diagnosed at admission (Paoli et al., 2018). Sepsis costs are even greater globally with the developing world at most risk. Altogether, sepsis is a major public health issue responsible for significant morbidity, mortality, and healthcare expenses. Early detection and antibiotic treatment of sepsis are critical for improving sepsis outcomes, where each hour of delayed treatment has been associated with roughly a 4-8% increase in mortality (Kumar et al., 2006; Seymour et al., 2017). To help address this problem, clinicians have proposed new definitions for sepsis (Singer et al., 2016), but the fundamental need to detect and treat sepsis early still remains, and basic questions about the limits of early detection remain unanswered. More on PhysioNet_Challenge.
3) Healthcare Time Series Data Synthesis:
Time series data is successive observations that represent measurements taken at equally spaced time intervals. Examples in healthcare include data from Electroencephalography (EEG) and activity recognition from accelerometers. Time series data collection methods are time-consuming and costly and such is the case when using gold-coated markers to track tumor movement in lung patients or the number for electrodes used to capture EEG data. Availability and access to large time-series datasets is limited yet it is an essential need for researchers seeking to build and test computational models used in the delivery of effective medical systems. The challenge here is to learn a data distribution that is not distorted while maintaining the time dependence and other important structural patterns. The ultimate aim is to explore unseen data and data patterns.
Posture Analysis, Alzheimer's, Autism, Hypertension, Chronic Pain, Home Monitoring Healthcare Service for Seniors