Automated rule-based decision systems in forensic toxicology using expert knowledge: Basic principles and practical applications

Abstract

This paper presents the basic principles and practical benefits of the application of expert systems (ES) and artificial intelligence (AI) to problem solving in forensic toxicology. We acknowledge the complexity and elegance of the theoretical substance and program algorithms of existing work in these disciplines, while simultaneously observing that many presentations of this material cloak the essential facts and concepts in unnecessary jargon and hyperbole. We attempt to remove the cloak without misrepresenting or oversimplifying the underlying structures. We first present a summary of the history, basic functions, technical fundamentals, and typical applications in three major categories of established ES/AI systems. We then assess the status of ES/AI in the forensic toxicology laboratory (FTL) with emphasis on potential applications. We conclude with an analysis of experiences with ESs in our laboratory where we have used an integrated expert system to reduce laboratory errors, detect internal inconsistencies in data, discover new substance abuse subpopulations, and reduce the frequency of sample reprocessing. We have minimized specimen processing time and instrument wear while maximizing technician efficiency and thus performing more tests for the same or reduced costs. © 1990 Oxford University Press.