Abstract
AbstractBackgroundThe “Body Mass Index” (BMI) or “Quetelet Index” is the most widely used tool to diagnose the degree of obesity. Anyone can calculate it, with no other tools than a scale and a measuring rod. However, its accuracy in predicting body fat percentage is low. The aim of this study is to find an alternative to BMI that is more reliable, accessible and easily applicable in daily clinical practice.MethodsUsing the K-means method (an unsupervised classification algorithm), we performed aclusteranalysis of the two phase angles obtained with bioimpedance analysis (BIA) of 641 women with different health status. BMI, age, diseases, treatments and any other data other than the phase angle values of the participants were not taken into account.ResultsTheclustersgenerated by the K-means algorithm do not coincide with the BMI categories, nor with the predetermined division of individuals into healthy and pathological.The K-means clustering algorithm identified new patterns that provide information on the greater or lesser predisposition of different individuals to suffer from diseases, taking as a reference their pathological peers in the samecluster.ConclusionsThe categories generated by the K-means algorithm based on the phase angles obtained by BIA classify individuals according to their health status independently of other variables such as age or BMI.
Publisher
Cold Spring Harbor Laboratory
Reference66 articles.
1. “Análisis de composición corporal y su uso en la práctica clínica en personas que viven con obesidad;Revista Médica Clínica Las Condes,2022
2. “Estimation of obesity levels based on computational intelligence;Informatics in Medicine Unlocked,2020
3. Accuracy of body mass index in diagnosing obesity in the adult general population
4. “Sensitivity and specificity of body mass index and main risk factors for cardiovascular disease in middle-income urban participants in Guanajuato, Mexico;Nutrición hospitalaria: Organo oficial de la Sociedad española de nutrición parenteral y enteral,2022
5. “BMI-BFMNU: A structural index linked to fat mass;Revista Española de Nutrición Humana y Dietética,2021