Bernard X W Liew1, Francisco M Kovacs2, David Rügamer3, Ana Royuela4. 1. School of Sport, Rehabilitation and Exercise Sciences, University of Essex, Colchester, Essex, UK. bl19622@essex.ac.uk. 2. Unidad de la Espalda Kovacs, Hospital Universitario HLA-Moncloa. University Hospital, Avenida de Menéndez Pelayo, 67, 28009, Madrid, Spain. 3. Department of Statistics, Ludwig-Maximilians-Universität München, München, Germany. 4. Biostatistics Unit. Hospital Puerta de Hierro, IDIPHISA, CIBERESP, REIDE, Madrid, Spain.
Abstract
PURPOSE: Prognostic models play an important clinical role in the clinical management of neck pain disorders. No study has compared the performance of modern machine learning (ML) techniques, against more traditional regression techniques, when developing prognostic models in individuals with neck pain. METHODS: A total of 3001 participants suffering from neck pain were included into a clinical registry database. Three dichotomous outcomes of a clinically meaningful improvement in neck pain, arm pain, and disability at 3 months follow-up were used. There were 26 predictors included, five numeric and 21 categorical. Seven modelling techniques were used (logistic regression, least absolute shrinkage and selection operator [LASSO], gradient boosting [Xgboost], K nearest neighbours [KNN], support vector machine [SVM], random forest [RF], and artificial neural networks [ANN]). The primary measure of model performance was the area under the receiver operator curve (AUC) of the validation set. RESULTS: The ML algorithm with the greatest AUC for predicting arm pain (AUC = 0.765), neck pain (AUC = 0.726), and disability (AUC = 0.703) was Xgboost. The improvement in classification AUC from stepwise logistic regression to the best performing machine learning algorithms was 0.081, 0.103, and 0.077 for predicting arm pain, neck pain, and disability, respectively. CONCLUSION: The improvement in prediction performance between ML and logistic regression methods in the present study, could be due to the potential greater nonlinearity between baseline predictors and clinical outcome. The benefit of machine learning in prognostic modelling may be dependent on factors like sample size, variable type, and disease investigated.
PURPOSE: Prognostic models play an important clinical role in the clinical management of neck pain disorders. No study has compared the performance of modern machine learning (ML) techniques, against more traditional regression techniques, when developing prognostic models in individuals with neck pain. METHODS: A total of 3001 participants suffering from neck pain were included into a clinical registry database. Three dichotomous outcomes of a clinically meaningful improvement in neck pain, arm pain, and disability at 3 months follow-up were used. There were 26 predictors included, five numeric and 21 categorical. Seven modelling techniques were used (logistic regression, least absolute shrinkage and selection operator [LASSO], gradient boosting [Xgboost], K nearest neighbours [KNN], support vector machine [SVM], random forest [RF], and artificial neural networks [ANN]). The primary measure of model performance was the area under the receiver operator curve (AUC) of the validation set. RESULTS: The ML algorithm with the greatest AUC for predicting arm pain (AUC = 0.765), neck pain (AUC = 0.726), and disability (AUC = 0.703) was Xgboost. The improvement in classification AUC from stepwise logistic regression to the best performing machine learning algorithms was 0.081, 0.103, and 0.077 for predicting arm pain, neck pain, and disability, respectively. CONCLUSION: The improvement in prediction performance between ML and logistic regression methods in the present study, could be due to the potential greater nonlinearity between baseline predictors and clinical outcome. The benefit of machine learning in prognostic modelling may be dependent on factors like sample size, variable type, and disease investigated.
Authors: Francisco M Kovacs; Jesús Seco; Ana Royuela; Sergio Melis; Carlos Sánchez; María J Díaz-Arribas; Marcelo Meli; Montserrat Núñez; María E Martínez-Rodríguez; Carmen Fernández; Mario Gestoso; Nicole Mufraggi; Jordi Moyá; Vicente Rodríguez-Pérez; Jon Torres-Unda; Natalia Burgos-Alonso; Inés Gago-Fernández; Víctor Abraira Journal: Clin J Pain Date: 2015-08 Impact factor: 3.442
Authors: Ana Royuela; Francisco M Kovacs; Carlos Campillo; Montserrat Casamitjana; Alfonso Muriel; Víctor Abraira Journal: Spine J Date: 2013-10-18 Impact factor: 4.166
Authors: Scott D Tagliaferri; Maia Angelova; Xiaohui Zhao; Patrick J Owen; Clint T Miller; Tim Wilkin; Daniel L Belavy Journal: NPJ Digit Med Date: 2020-07-09
Authors: Francisco M Kovacs; Jesús Seco-Calvo; Borja M Fernández-Félix; Javier Zamora; Ana Royuela; Alfonso Muriel Journal: BMC Musculoskelet Disord Date: 2019-12-26 Impact factor: 2.362
Authors: Harry Hemingway; Peter Croft; Pablo Perel; Jill A Hayden; Keith Abrams; Adam Timmis; Andrew Briggs; Ruzan Udumyan; Karel G M Moons; Ewout W Steyerberg; Ian Roberts; Sara Schroter; Douglas G Altman; Richard D Riley Journal: BMJ Date: 2013-02-05
Authors: Francisco M Kovacs; Joan Bagó; Ana Royuela; Jesús Seco; Sergio Giménez; Alfonso Muriel; Víctor Abraira; José Luis Martín; José Luis Peña; Mario Gestoso; Nicole Mufraggi; Montserrat Núñez; Josep Corcoll; Ignacio Gómez-Ochoa; Ma José Ramírez; Eva Calvo; Ma Dolores Castillo; David Martí; Salvador Fuster; Carmen Fernández; Nuria Gimeno; Alejandro Carballo; Alvaro Milán; Dolores Vázquez; Montserrat Cañellas; Ricardo Blanco; Pilar Brieva; Ma Trinidad Rueda; Luis Alvarez; María Teresa Gil Del Real; Joaquín Ayerbe; Luis González; Leovigildo Ginel; Mariano Ortega; Miryam Bernal; Gonzalo Bolado; Anna Vidal; Ana Ausín; Domingo Ramón; María Antonia Mir; Miquel Tomás; Javier Zamora; Alejandra Cano Journal: BMC Musculoskelet Disord Date: 2008-04-09 Impact factor: 2.362