Positions

Selected Publications

Academic Article

Year Title Altmetric
2021 A Bayesian approach for estimating age-adjusted rates for low-prevalence diseases over space and timeStatistics in Medicine.  40:2922-2938. 2021
2021 Displaying survival of patient groups defined by covariate paths: Extensions of the Kaplan-Meier estimatorStatistics in Medicine.  40:2024-2036. 2021
2021 Predicting an optimal composite outcome variable for Huntington's disease clinical trialsJournal of Applied Statistics.  48:1339-1348. 2021
2018 Multicentre validation of a sepsis prediction algorithm using only vital sign data in the emergency department, general ward and ICUBMJ Open.  8. 2018
2017 Cost and mortality impact of an algorithm-driven sepsis prediction system 2017
2017 Using Transfer Learning for Improved Mortality Prediction in a Data-Scarce Hospital Setting.Biomedical Informatics Insights.  9:1178222617712994. 2017
2016 Using electronic health record collected clinical variables to predict medical intensive care unit mortalityAnnals of medicine and surgery (2012).  11:52-57. 2016
2016 A computational approach to mortality prediction of alcohol use disorder inpatientsComputers in Biology and Medicine.  75:74-79. 2016
2016 A computational approach to early sepsis detectionComputers in Biology and Medicine.  74:69-73. 2016
2016 Prediction of sepsis in the intensive care unit with minimal electronic health record data: A machine learning approachJMIR Medical Informatics.  4. 2016
2016 High-performance detection and early prediction of septic shock for alcohol-use disorder patientsAnnals of medicine and surgery (2012).  8:50-55. 2016
2015 Determining the top all-time college coaches through Markov chain-based rank aggregation 2015

Research Overview

  • Dr. Melissa Jay Smith's statistical research interests are in the development of Bayesian methods for disease mapping and mediation analysis. The goal of disease mapping is to create reliable model-based estimates of disease risk in small areas such as ZIP codes or counties. By visualizing these estimates on a map, public health professionals and policymakers can determine where levels of a disease are high and low and effectively allocate limited resources or plan prevention efforts. Her recent work focuses on a new Bayesian method for estimating age-adjusted incidence and mortality rates when there are excess zeros in the dataset due to the low-prevalence of a disease and small population sizes in rural areas.

    Dr. Smith also conducts research in the area of mediation analysis. The purpose of a mediation analysis is to better understand the mechanism through which an exposure affects a health outcome. In epidemiological studies, mediation analyses are also frequently used to understand why disease incidence or mortality differs between groups. Dr. Smith develops and evaluates statistical methods for conducting these types of analyses.

    In addition to her statistical research, Dr. Smith is interested in interdisciplinary studies that use electronic health records for health outcome prediction and comparative effectiveness research. Previously, she worked at a machine learning start-up company focused on predicting sepsis and patient decompensation in the hospital using data from patients’ charts. In this position, she managed multiple databases of electronic health records, created statistical tables and figures, and drafted publications and case studies.

    Keywords: Bayesian statistics, spatial statistics, disease mapping, mediation analysis, health disparities, electronic health records.
  • Education And Training

  • Doctor of Philosophy in Biostatistics, University of Iowa 2022
  • Master of Sciences or Mathematics in Biostatistics, University of Iowa 2019
  • Bachelor of Arts in Mathematics, Colorado College 2016
  • Full Name

  • Melissa Smith