By Braga, A.C.; Carneiro, P.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Abstract. Multivariable prediction models are being developed, validated, updated, and implemented in different domains, with the aim of assisting professionals in estimating probabilities and potentially influence their decision making. The main goal of this work was to develop and validate a logistic regression model that would predict the risk of musculoskeletal complaints involving the lumbar region in nurses working at the Health Centres of Northern Portugal and providing home-based care. The main methodology used was a questionnaire developed in electronic format, which was based on the Standardized Nordic Questionnaire for the analysis of musculoskeletal symptoms. Internal and external validation methods were used, and their performances were compared with the ROC (Receiver Operating Characteristic) methodology.