Training
Home Variable selection in predictive models with anthropometric measurements
SCIENTIFIC SESSION
Variable selection in predictive models with anthropometric measurements
Speaker: Sara Rodríguez Pastoriza
With a degree in Mathematics and Statistics, currently studying for a Master’s in Statistical Techniques at the University of Santiago de Compostela. Martín Fernández Pérez, Biology graduate and Master’s degree holder in statistical techniques and operational research from USC. Both currently fellows at the Galicia Sur Health Research Institute, affiliated with the I-Saúde research group.
Brief Description of the Session
Body composition in childhood and adolescence shows major changes over short periods of time. Obesity is an excess of body mass and fat; in addition to being an increasingly prevalent condition, it is associated with significant comorbidities and a reduced quality of life in adulthood. To measure it, bioelectrical impedance analysis is used, a technique not available in primary care.
Therefore, in order to estimate fat mass in a simple, acceptable and easy way in day-to-day clinical practice, a new formula is presented which involves various anthropometric measurements, as well as a free online calculator to classify the risk of overweight and/or obesity in adolescents.
We present a variable selection algorithm in Generalised Additive Models based on the cross-validation technique to optimise AUC and its diagnostic utility.
The selected predictive models were deployed in a web-based calculator for use in clinical practice.
93 482 46 94
671 511 120
info@ricapps.es
