Computational Estimation and Prediction in Public Health

<p>Using computational approaches utilizing large datasets to investigate public health</p><p>information is an important mechanism for institutions seeking to identify strategies</p><p>for improving public health. The art in computational approaches for example</p><p>in health research is managing the trade-offs between the two perspectives:</p><p>first inference and s econd p rediction. Many techniques from statistical methods</p><p>(SM) and machine learning (ML) may in principle be used for both perspectives.</p><p>However SM has a well established focus on inference by building probabilistic</p><p>models which allows us to determine a quantitative measure of confidence about</p><p>the magnitude of the effect. Simulation-based validation approaches can be used</p><p>in conjunction with SM to explicitly verify assumptions and redefine t he specified</p><p>model if n ecessary. On the other hand ML uses general-purpose algorithms</p><p>to find p atterns t hat b est p redict t he o utcome and makes minimal assumptions</p><p>about the data-generating process; and may be more effective in a number of situations.</p><p>My work employs both SM- and ML- based computational approaches to</p><p>investigate particular public health problems. Chapter One provides philosophical</p><p>background and compares the application of the two approaches in public health.</p><p>Chapter Two describes and implements penalized Cox proportional hazard models</p><p>for time-varying covariates time-to-event data. Chapter Three applies traditional</p><p>survival models and machine learning algorithms to predict survival times of cancer</p><p>patients while incorporating the information about the time-varying covariates.</p><p>Chapter Four discusses and implements various approaches for computing predictions</p><p>and effects for generalized linear (mixed) models. Finally Chapter Five</p><p>implements and compares various statistical models for handling univariate and</p><p>multivariate binary outcomes for water sanitation and hygiene (WaSH) data.</p>