Artificial Intelligence in Forensic Toxicology: A Systematic Review

Artificial Intelligence (AI) is increasingly transforming the analytical and investigative landscape of forensic toxicology. Traditional toxicological workflows rely on gas/liquid chromatography-mass spectrometry (GC-MS/LC-MS), manual interpretation, and expert judgment face mounting challenges from data complexity, novel psychoactive substances (NPS), and throughput demands. This systematic review synthesizes evidence from 17 primary studies (2019–2026) to map how AI and machine learning (ML) are integrated across the forensic toxicology pipeline, spanning novel psychoactive substance identification, predictive toxicology and risk assessment, postmortem interval (PMI) estimation, and laboratory workflow automation. A structured literature search was conducted across PubMed, Scopus, Web of Science, and Google Scholar using predefined MeSH and keyword terms. Studies were screened using PRISMA guidelines. Seventeen studies were included based on defined eligibility criteria covering original research, systematic reviews, and comprehensive narrative reviews on AI applications in forensic and analytical toxicology. Across domains, AI and ML methods including deep generative models, convolutional and artificial neural networks, random forests, and explainable AI frameworks has consistently improved analytical sensitivity, processing speed, and detection of previously unidentifiable substances, while highlighting persistent challenges around dataset standardization, model generalizability, and forensic/legal admissibility.