The purpose of this study has been to develop a multivariate model for predicting body fat mass in women by using the technique of infrared (IR) thermography. Sixty-nine healthy women, aged from 16 to 29, were investigated by using a body composition analyser and IR thermographic temperature measurement. The correlation analysis was performed to reveal the problem of multicollinearity. The technique of principal component analysis (PCA) was applied in order to reduce the number of variables. Both the total fat mass and the fat mass in the torso were accepted as the dependent variables. The individual scores were used as independent variables on each component after applying the orthogonal rotation. Two datasets were analysed: the full dataset with anthropometric characteristics (age, body mass, body length) and without anthropometric characteristics. The stepwise model meeting the Akaike information criterion (AIC) was selected to estimate the relative quality of all models. The models obtained on the full dataset were able to explain 73.9% of the fat mass in the torso and 70.4% of the total fat mass. Respectively, the models based on the reduced dataset explained 52.5% of the fat mass in the torso and 51.5% of the total fat mass.

这项研究的目的是通过使用红外（IR）热成像技术来开发预测女性体内脂肪量的多元模型。通过使用人体成分分析仪和红外热像仪温度测量对69名年龄在16至29岁之间的健康女性进行了调查。进行相关分析以揭示多重共线性问题。为了减少变量数，应用了主成分分析（PCA）技术。总脂肪量和躯干中的脂肪量均被接受为因变量。在应用正交旋转后，将各个分数用作每个组件上的自变量。分析了两个数据集：具有人体测量特征（年龄，体重，身体长度）且没有人体测量特征。选择满足Akaike信息标准（AIC）的逐步模型来估计所有模型的相对质量。在完整数据集上获得的模型能够解释躯干中73.9％的脂肪量和总脂肪量的70.4％。基于精简数据集的模型分别解释了躯干中52.5％的脂肪和总脂肪中的51.5％。