Doctoral Dissertations
Date of Award
12-2024
Degree Type
Dissertation
Degree Name
Doctor of Philosophy
Major
Data Science and Engineering
Major Professor
Audris Mockus
Committee Members
Audris Mockus, Russell Zaretzki, Dawnie W. Steadman, Amir Sadovnik
Abstract
Determining the postmortem interval (PMI)—the time elapsed between death and the discovery of a body—is a fundamental challenge in forensic anthropology. Accurate PMI estimation is essential for narrowing down the list of potential decedents, facilitating the identification process, and assisting law enforcement in eliminating suspects or verifying witness statements in cases of homicide. Additionally, knowledge of the PMI provides critical insights into the environmental and biological factors that have influenced the decomposition process over time, thereby enabling a more comprehensive analysis of the remains.
The Forensic Anthropology Center (FAC) at the University of Tennessee, Knoxville (UTK), is home to the Anthropology Research Facility (ARF), where human decomposition has been studied since 1981. Since 2012, photographic documentation of approximately 100 donors per year has resulted in a repository of over one million images from more than 800 donors. This extensive photographic collection, presents an unparalleled resource for forensic research.
In 2016, the Departments of Electrical Engineering and Computer Science and Anthropology at UTK embarked on a collaborative project funded by the National Institute of Justice (NIJ). The project's primary objectives are to make this vast photographic collection accessible and useful for forensic research and casework and to develop advanced artificial intelligence (AI) tools specifically for forensic purposes, with a primary focus being PMI estimation.
This dissertation contributes to these ongoing efforts by: (1) continuing to develop and maintain a collaborative tool designed to organize, visualize, analyze, and curate the large collection of human decomposition photographs, thus enhancing their usability in forensic research; (2) creating a method to label stages of decay when labeled data is limited and subsequently devising an AI-driven approach to automate the identification of these stages; and (3) evaluating existing PMI estimation methods using this extensive dataset, providing a more comprehensive and statistically significant analysis than earlier studies. Through these contributions, this work aims to advance the field of forensic anthropology by leveraging AI and large-scale data to improve the accuracy and reliability of PMI estimation, thereby enhancing the overall effectiveness of forensic investigations.
Recommended Citation
Nau, Anna-Maria, "Applications of Machine Learning in Forensic Investigations of Human Decomposition. " PhD diss., University of Tennessee, 2024.
https://trace.tennessee.edu/utk_graddiss/11377
Included in
Artificial Intelligence and Robotics Commons, Data Science Commons, Forensic Science and Technology Commons