Modelo de predicción Multivariable de Resultados de Artroplastia Total de Rodilla

Total knee arthroplasty (TKA) is a surgical procedure applied as a common solution to overcome limitations produced by advanced stages of severe gonarthrosis. The procedure has high prevalence, high associated costs, and is considered to be cost-effective. Rehabilitation is essential to optimize outcomes. However, in clinical practice, the length of rehabilitation for each patient may be highly variable, and the programmed times may lack the necessary objectivity. Current limitation of resources and increasing prevalence make essential to generate strategies to optimize surgical results, so that the use of resources of the health system is efficient without detriment to the patient's benefit. For this purpose, objective and pragmatic information must be available, and should be based on scientific evidence in order to assist in making clinical decisions. Indeed, a number of demographic, biomedical and psychosocial factors have been identified as predictors of TKA results (i.e weight, age, expectations...). Some of them have been associated with the need for hospital resources after surgery. However, most researches base their predictions in retrospective studies, which are limited in the type of variables that can be used (clinic history), quality of registries, and limitations of retrospective designs. On the other hand, most of prospective researches base their predictions in a limited number of outcomes. To overcome this limitations, this project has been designed as a prospective observational study with two observations of each patient.

• The primary goal is to implement a multi-variable prediction model of TKA outcome, so that the procedure become optimal in two aspects : patient recovery (social and economic benefit) and use of health system resources (economic benefit). The implementation requires a processing of the information sampled through various algorithms and innovative data processing in this field, based on data mining and machine learning techniques. This will be used in search of the model with the greatest predictive capacity.
• As a secondary objective, information extracted from patients both in the final stages of the condition, and in the medium term after the intervention will allow to study the functional and psychosocial reality of the subjects with knee osteoarthritis.