Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/133226
Title: Registres de salut digitals: Tractament de dades i construcció de models de predicció de malaltia
Author: Cros Roura, Sílvia
Tutor: Perez-Alvarez, Nuria  
Vegas, Esteban  
Others: Maceira, Marc  
Abstract: The main objectives pursued by this TFM were how to deal with data derived from Electronic Health Records and to simulate a prediction study for classifying patients according to the risk of developing rheumatoid arthritis using this data. For that, a 100.000 virtual patients' dataset was used. This cohort was adapted and reduced to 28.572 patients; half diagnosed with RA. A statistical analysis was performed, with two differentiated parts: the first one consisted in the identification of risk factors associated with the disease, recognizing hypoalbuminemia, proteinemia, anaemia, leucocytosis, thrombocytosis, the feminine gender, and an age over than 45 years as correlated variables. The second part used these factors to build and execute the different selected models, which were: Logistic multiple regression, Naïve Bayes algorithm, Random Forest, SVM and ANN. Their performance was evaluated by means of different parameters, such as: ROC curves, accuracy, and AUC. Excluding Naïve Bayes, all the other models showed a good performance, hence, all are considered acceptable to be used in classification problems based in numeric and categoric predictors. Nevertheless, it must be taken into account that the work was done with simulated data, therefore, the conclusions are not comparably to the real patients.
Keywords: risk factors
prediction model
rheumatoid arthritis
clinical trials
health digital records
Document type: info:eu-repo/semantics/masterThesis
Issue Date: 28-Jun-2021
Publication license: http://creativecommons.org/licenses/by-nc-nd/3.0/es/  
Appears in Collections:Trabajos finales de carrera, trabajos de investigación, etc.

Files in This Item:
File Description SizeFormat 
scrosroTFM0621memòria.pdfMemòria del TFM2,07 MBAdobe PDFThumbnail
View/Open