Development and comparison of forensic interval age prediction models by statistical and machine learning methods based on the methylation rates of ELOVL2 in blood DNA
Takayuki Yamagishi, Wataru Sakurai, Ken Watanabe, Kochi Toyomane, Tomoko Akutsu
{"title":"Development and comparison of forensic interval age prediction models by statistical and machine learning methods based on the methylation rates of ELOVL2 in blood DNA","authors":"Takayuki Yamagishi, Wataru Sakurai, Ken Watanabe, Kochi Toyomane, Tomoko Akutsu","doi":"10.1016/j.fsigen.2023.103004","DOIUrl":null,"url":null,"abstract":"<div><p><span><span><span>Age estimation can be useful information for narrowing down candidates of unidentified donors in criminal investigations. Various age estimation models based on DNA methylation biomarkers have been developed for forensic usage in the past decade. However, many of these models using ordinary </span>least squares<span> regression cannot generate an appropriate estimation due to the deterioration in prediction accuracy caused by an increased prediction error in older age groups. In the present study, to address this problem, we developed age estimation models that set an appropriate prediction interval for all age groups by two approaches: a statistical method using quantile regression (QR) and a machine learning method using an artificial </span></span>neural network<span> (ANN). Methylation datasets (</span></span><em>n</em> = 1280, age 0–91 years) of the promoter for the gene encoding <span><em>ELOVL fatty acid </em><em>elongase</em><em> 2</em></span> were used to develop the QR and ANN models. By validation using several test datasets, both models were shown to enlarge prediction intervals in accordance with aging and have a high level of correct prediction (>90 %) for older age groups. The QR and ANN models also generated a point age prediction with high accuracy. The ANN model enabled a prediction with a mean absolute error (MAE) of 5.3 years and root mean square error (RMSE) of 7.3 years for the test dataset (<em>n</em> = 549), which were comparable to those of the QR model (MAE = 5.6 years, RMSE = 7.8 years). Their applicability to casework was also confirmed using bloodstain samples stored for various periods of time (1–14 years), indicating the stability of the models for aged bloodstain samples. From these results, it was considered that the proposed models can provide more useful and effective age estimation in forensic settings.</p></div>","PeriodicalId":50435,"journal":{"name":"Forensic Science International-Genetics","volume":null,"pages":null},"PeriodicalIF":3.2000,"publicationDate":"2023-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Forensic Science International-Genetics","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1872497323001795","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"GENETICS & HEREDITY","Score":null,"Total":0}
引用次数: 0
Abstract
Age estimation can be useful information for narrowing down candidates of unidentified donors in criminal investigations. Various age estimation models based on DNA methylation biomarkers have been developed for forensic usage in the past decade. However, many of these models using ordinary least squares regression cannot generate an appropriate estimation due to the deterioration in prediction accuracy caused by an increased prediction error in older age groups. In the present study, to address this problem, we developed age estimation models that set an appropriate prediction interval for all age groups by two approaches: a statistical method using quantile regression (QR) and a machine learning method using an artificial neural network (ANN). Methylation datasets (n = 1280, age 0–91 years) of the promoter for the gene encoding ELOVL fatty acid elongase 2 were used to develop the QR and ANN models. By validation using several test datasets, both models were shown to enlarge prediction intervals in accordance with aging and have a high level of correct prediction (>90 %) for older age groups. The QR and ANN models also generated a point age prediction with high accuracy. The ANN model enabled a prediction with a mean absolute error (MAE) of 5.3 years and root mean square error (RMSE) of 7.3 years for the test dataset (n = 549), which were comparable to those of the QR model (MAE = 5.6 years, RMSE = 7.8 years). Their applicability to casework was also confirmed using bloodstain samples stored for various periods of time (1–14 years), indicating the stability of the models for aged bloodstain samples. From these results, it was considered that the proposed models can provide more useful and effective age estimation in forensic settings.
期刊介绍:
Forensic Science International: Genetics is the premier journal in the field of Forensic Genetics. This branch of Forensic Science can be defined as the application of genetics to human and non-human material (in the sense of a science with the purpose of studying inherited characteristics for the analysis of inter- and intra-specific variations in populations) for the resolution of legal conflicts.
The scope of the journal includes:
Forensic applications of human polymorphism.
Testing of paternity and other family relationships, immigration cases, typing of biological stains and tissues from criminal casework, identification of human remains by DNA testing methodologies.
Description of human polymorphisms of forensic interest, with special interest in DNA polymorphisms.
Autosomal DNA polymorphisms, mini- and microsatellites (or short tandem repeats, STRs), single nucleotide polymorphisms (SNPs), X and Y chromosome polymorphisms, mtDNA polymorphisms, and any other type of DNA variation with potential forensic applications.
Non-human DNA polymorphisms for crime scene investigation.
Population genetics of human polymorphisms of forensic interest.
Population data, especially from DNA polymorphisms of interest for the solution of forensic problems.
DNA typing methodologies and strategies.
Biostatistical methods in forensic genetics.
Evaluation of DNA evidence in forensic problems (such as paternity or immigration cases, criminal casework, identification), classical and new statistical approaches.
Standards in forensic genetics.
Recommendations of regulatory bodies concerning methods, markers, interpretation or strategies or proposals for procedural or technical standards.
Quality control.
Quality control and quality assurance strategies, proficiency testing for DNA typing methodologies.
Criminal DNA databases.
Technical, legal and statistical issues.
General ethical and legal issues related to forensic genetics.