Scholarlink

<p>Accurate estimation of the postmortem interval (PMI) is essential in forensic medicine for reconstructing the timeline and circumstances of death. Artificial intelligence (AI) has emerged in recent years as a promising tool to enhance this estimation through the analysis of complex biological data. This study aims to conduct a systematic review of recent advances in AI applied to PMI estimation, complemented by a meta-analysis assessing the predictive performance of commonly used AI models such as neural networks, ensemble models, and random forest, using the area under the curve (AUC) as the primary metric. A literature search was conducted across PubMed, Scopus, and Google Scholar for the period 2015–2025, identifying 16 eligible studies. The analyzed models integrated microbiological, proteomic, imaging, and spectroscopic data, achieving over 90% accuracy in several studies. The meta-analysis, based on five studies with comparable data, yielded a combined AUC of 0.94 (95% CI ((Confidence Interval): 0.81–1.08), with no significant heterogeneity or publication bias. These findings highlight the strong potential of AI—particularly when combined with multi-omics approaches—as a precise and robust method for PMI estimation. This approach addresses several limitations of traditional forensic methods, although certain technical and implementation challenges remain to be resolved.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Application of artificial intelligence in the determination of the postmortem interval: Systematic review of the literature and metaanalysis

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月31日

查看全文
  • Lidaray Cuba-Gutierrez,
  • Marina Invernón-Monedero,
  • Eduardo Osuna,
  • Diana Hernández-Romero

摘要

Accurate estimation of the postmortem interval (PMI) is essential in forensic medicine for reconstructing the timeline and circumstances of death. Artificial intelligence (AI) has emerged in recent years as a promising tool to enhance this estimation through the analysis of complex biological data. This study aims to conduct a systematic review of recent advances in AI applied to PMI estimation, complemented by a meta-analysis assessing the predictive performance of commonly used AI models such as neural networks, ensemble models, and random forest, using the area under the curve (AUC) as the primary metric. A literature search was conducted across PubMed, Scopus, and Google Scholar for the period 2015–2025, identifying 16 eligible studies. The analyzed models integrated microbiological, proteomic, imaging, and spectroscopic data, achieving over 90% accuracy in several studies. The meta-analysis, based on five studies with comparable data, yielded a combined AUC of 0.94 (95% CI ((Confidence Interval): 0.81–1.08), with no significant heterogeneity or publication bias. These findings highlight the strong potential of AI—particularly when combined with multi-omics approaches—as a precise and robust method for PMI estimation. This approach addresses several limitations of traditional forensic methods, although certain technical and implementation challenges remain to be resolved.