LEARNING TO GENERATE INDIVIDUAL DATA SEQUENCE FROM POPULATION STATISTICS USING DYNAMIC BAYESIAN NETWORKS

Abstract

Data collection rose exponentially with the dawn of the 21st Century, However the most important data to humans, individual health data, is difficult to get approved for public research, as medical history is very sensitive to be distributed. The only available public data which can be retrieved from institutions like the Centre for Disease Control (CDC), World Health Organization (WHO), National Health Interview Survey (NHIS), etc. largely only contain population statistics for different attributes of a person.What we propose here is a generative model which would learn to create data sequences for a population, each sequence mimicking an individual person’s behavior, such that the set of generated sequences represents this entire population by matching the available population statistics using Dynamic Bayesian Networks. The data would contain a population in which each person will have their exercise, injury and illness data over time. Various factors are interlinked within and between time slices, e.g. the amount of exercise a person does at time t, depends on factors [Age, Health-Status, Person Type] at time t, and exercise done at time t-1. Although the data generated by the learning model is not real, it should be closer to it, supporting algorithm development and initial testing.