New perspectives of forensic pathology through machine learning approach on autopsy data: a pilot study

• Published in: Clinica terapeutica Vol. 175; no. Suppl 1(4); p. 23
• Main Authors: Cordasco, F, Sacco, M A, Gualtieri, S, Tarzia, P, Pulpito, G, Aquila, I
• Format: Journal Article
• Language: English
• Published: Italy 01.07.2024
• Subjects: Adult; Aged; Autopsy - methods; Cause of Death; Female; Forensic Pathology - methods; Humans; Machine Learning; Male; Middle Aged; Organ Size; Pilot Projects; artificial intelligence; autopsy; cause of death; forensic pathology; machine learning; statistical analysis
• ISSN: 1972-6007
• DOI: 10.7417/CT.2024.5078

The analysis, interpretation and storage of information is entrusted to the individual expert, who bases his judgments on the knowledge resulting from the experience. The aim of this experimental study is to analyse and introduce a new line of research applicable to forensic pathology, based on the use of artificial intelligence techniques as a possible tool for data collection and analysis. The sample analysed is represented by judicial autop-sies performed at the University of Catanzaro from 01/01/2020 to 31/12/2021. For each case were performed: study of medical records; autopsy; histological examinations; toxicological analysis of blood samples. Continuous variables were presented as means ± standard deviations, and categorical variables were expressed as percentages. A random forest regression model was conducted, as a machine learning approach, to estimate the importance of individual solid organ weight variables in predicting cause of death. This study aimed to evaluate autopsy data to aid in the description and study of forensic cases, using a machine learning approach. To date, this study appears to be the first to evaluate the weight of organs in predicting a cause of death. Artificial intelligence techniques are an optimal solution in solving forensic dilemmas. The results of this study demonstrate that routine data can be submitted using machine learning techniques in order to identify key elements of procedures that provide more information in relation to the predic-tion of cause of death.