Abstract:The prediction of the chronological age of a deceased individual at time of death can provide important information in case of unidentified bodies. The methodological possibilities in these cases depend on the availability of tissues, whereby bones are preserved for a long time due to their mineralization under normal environmental conditions. Age-dependent changes in DNA methylation (DNAm) as well as the accumulation of pentosidine (Pen) and D-aspartic acid (D-Asp) could be useful molecular markers for age prediction. A combination of such molecular clocks into one age prediction model seems favorable to minimize inter- and intra-individual variation. We therefore developed (I) age prediction models based on the three molecular clocks, (II) examined the improvement of age prediction by combination, and (III) investigated if samples with signs of decomposition can also be examined using these three molecular clocks. Skull bone from deceased individuals was collected to obtain a training dataset (n = 86), and two independent test sets (without signs of decomposition: n = 44, with signs of decomposition: n = 48). DNAm of 6 CpG sites in ELOVL2, KLF14, PDE4C, RPA2, TRIM59 and ZYG11A was analyzed using massive parallel sequencing (MPS). The D-Asp and Pen contents were analyzed by high performance liquid chromatography (HPLC). Age prediction models based on ridge regression were developed resulting in mean absolute errors (MAEs)/root mean square errors (RMSE) of 5.5years /6.6 years (DNAm), 7.7 years /9.3 years (Pen) and 11.7 years /14.6 years (D-Asp) in the test set. Unsurprisingly, a general lower accuracy for the DNAm, D-Asp, and Pen models was observed in samples from decomposed bodies (MAE: 7.4-11.8 years, RMSE: 10.4-15.4 years). This reduced accuracy could be caused by multiple factors with different impact on each molecular clock. To acknowledge general changes due to decomposition, a pilot model for a possible age prediction based on the decomposed samples as training set improved the accuracy evaluated by leave-one-out-cross validation (MAE: 6.6-12 years, RMSE: 8.1-15.9 years). The combination of all three molecular age clocks did reveal comparable MAE and RMSE results to the pure analysis of the DNA methylation for the test set without signs of decomposition. However, an improvement by the combination of all three clocks was possible for the decomposed samples, reducing especially the deviation in case of outliers in samples with very high decomposition and low DNA content. The results demonstrate the general potential in a combined analysis of different molecular clocks in specific cases.

Combination of several predictors for forensic age estimation

Age estimation based on different molecular clocks in several tissues and a multivariate approach: an explorative study

A probability model for estimating age in young individuals relative to key legal thresholds: 15, 18 or 21-year

FORENSIC DNA PHENOTYPING FOR CRIMINAL IDENTIFICATION

Age Classification in Forensic Medicine Using Machine Learning Techniques

A novel strategy for forensic age prediction by DNA methylation and support vector regression model

Research Progress of Age Estimation Based on the Demirjian's Method

Accuracy of geometric morphometrics for age estimation using frontal face photographs of children and adolescents: A promising method for forensic practice

Forensic Age Estimation through a DNA Methylation-Based Age Prediction Model in the Italian Population: A Pilot Study

Forensic Age Estimation Based on the Pigmentation in the Costal Cartilage from Human Mortal Remains.

Progress in Age Estimation Based on DNA Methylation

Multivariate cumulative probit for age estimation using ordinal categorical data

Molecular age prediction using skull bone samples from individuals with and without signs of decomposition: a multivariate approach combining analysis of posttranslational protein modifications and DNA methylation

Estimating sex and age for forensic applications using machine learning based on facial measurements from frontal cephalometric landmarks

Collaborative exercise: analysis of age estimation using a QIAGEN protocol and the PyroMark Q48 platform

Evaluation of DNA Methylation-Based Age-Prediction Models from Saliva and Buccal Swab Samples Using Pyrosequencing Data

Age-related DNA methylation analysis for forensic age estimation using post-mortem blood samples from Japanese individuals

Contribution of Anatomy in Forensic Age Evaluation: A Systematic Review

Considerations for age estimation accuracy: Method‐derived outcomes and practitioner interpretations

Forensic age estimation at the University Center of Legal Medicine Lausanne-Geneva: a retrospective study over 12 years

COMBI: Artificial Intelligence for Computer-Based Forensic Analysis of Persons