Determination of a person’s age can be an important factor in forensic investigation. DNA methylation (DNAm) is a well-known factor signifying change during the aging process but also necessary for the development of mammals. Several studies reported that DNAm can be used as an important marker in predicting the age of a human. This study is carried out to develop the age prediction model using three different regression methods. Multiple linear regression, Support vector regression, and Random forest regression methods are applied using a set of four highly age-correlated CpG sites. For 180 blood samples having age between 2 and 87 years, the mean absolute deviation (MAD) for multiple linear regression method is 8.43 years, for support vector regression is 7.86 years and for random forest regression method is 8.25 years. Further, these models are tested on five different age-groups. The average MAD for multiple linear regression, support vector regression and random forest regression are 3.46, 3.44 and 3.56, respectively. Support vector regression gave the highest accuracy for combined samples as well as for 5 different age groups. It has been concluded from the results that support vector regression is a reliable method for human age prediction.

中文翻译：

---

使用DNA甲基化和回归方法预测人类年龄

确定一个人的年龄可能是法医调查的重要因素。DNA甲基化（DNAm）是众所周知的因子，表示衰老过程中发生变化，但对于哺乳动物的发育也是必需的。几项研究报告说，DNAm可用作预测人类年龄的重要标志。进行这项研究以使用三种不同的回归方法开发年龄预测模型。使用一组四个与年龄高度相关的CpG位点应用了多元线性回归，支持向量回归和随机森林回归方法。对于年龄在2至87岁之间的180个血液样本，多元线性回归方法的平均绝对偏差（MAD）为8.43年，支持向量回归的平均绝对偏差（MAD）为7.86年，而随机森林回归方法的平均绝对偏差为8.25年。进一步，这些模型在五个不同年龄组进行了测试。多元线性回归，支持向量回归和随机森林回归的平均MAD分别为3.46、3.44和3.56。支持向量回归为合并样本以及5个不同年龄组的样本提供了最高的准确性。从结果可以得出结论，支持向量回归是用于人类年龄预测的可靠方法。