Sex Determination from the Metacarpals in Macaac mulatta using Logistic Regression

Number of views： 10

分享到:

摘要:

对猕猴掌骨变量的性差进行逻辑回归分析,用ROC分析来比较掌骨变量、左右侧掌骨和5根掌骨的性差大小.成年掌骨标本44例（14雄,30雌）.数据分析采用SPSS20.0软件.选择掌骨的7个变量.采用逐步逻辑回归分析选择最佳变量.结果表明：掌骨长度是最好的性差指示变量,然后是掌骨头和掌骨底变量,最小的是掌骨骨干宽和高.总体性别正确判别率是71.4%~100.0%.ROC分析显示左右侧掌骨性差差异很小,ROC曲线下面积（AUC）分别是0.969和0.978.逻辑回归结果是MC2、MC4和MC5判别率较高（90.0%~94.9%）;MC1和MC3判别率较低（85.8%~87.5%）.结果提示：猕猴掌骨性差显著.掌骨性差模式可能与人类不同.掌骨变量的性差特征反映了遗传与后天行为因素的交互作用.

Logistic regression was used in order to predict sex based on values of predictor variables of the metacarpal. ROC analysis was used to assess sexing performance at several different measurement values, different sides of the body throughout the distribution and which metacarpals would produce the greatest accuracy. The sample selected for this study con- sists of 44 skeletons of adult Macaac mulatta （14 males, 30 females） from Taihang Mountains. Data was analyzed using SPSS 20.0 version. Seven variables were taken on each metacarpals. Stepwise logistic regression model were performed using all sev- en variables to choose the best variable. Results.. In the models the best variaigle for the metacarpals is LG followed by HH, HW, BH and BW Logistic regression models were formed to estimate sex, ranging from 71.4-100.0 for the pooled indi- viduals being correctly classified. There were very small bilateral differences by using ROC analysis, and the AUC values were similar between two sides, with 0. 969 and 0. 978 for left and right metacarpals, respectively. For five metacarpals of both hands in this study, the percentage of correct classification was higher for MC2, MC4 and MC5 （90.0 %-94.9 ） and lower MC1 and MC3 （85.8-87.5）, respectively. In general, results from this study suggest that the measurements of metacar- pals appear to be good discriminators of sex. These findings suggest that the pattern of sexual dimorphism for metacarpals may be different from that in human.

作者:

胡凤霞胡海洋田华香赵哲苏瑞平赵晓进

机构地区:

betway官方app 水产学院子 betway官方app 生命科学学院

出处:

《betway官方app 学报：自然科学版》 CAS 北大核心 2015年第3期125-128,177,共5页

基金:

河南省基础与前沿技术研究计划项目(132300410136)

关键词:

猕猴掌骨性差逻辑回归 ROC分析

Macaca mulatta metacarpal sex dimorphism logistic regression l ROC analysis

分类号: