Classification methods applied to the diagnosis of pediatric patients with familial hypercholesterolemia: comparison of Simon Broome criteria with logistic regression and modified decision tree models

Albuquerque, João; Antunes, Marília; Alves, Ana Catarina; Medeiros, Ana Margarida; Bourbon, Mafalda

http://hdl.handle.net/10400.18/6658

Use this identifier to reference this record.

Name:	Description:	Size:	Format:
Classification Methods applied to the Diagnosis of Pediatric Patients with Familial.pdf		748.01 KB	Adobe PDF	Download

Send Feedback

Authors

Albuquerque, João

Antunes, Marília

Alves, Ana Catarina

Medeiros, Ana Margarida

Bourbon, Mafalda

Abstract(s)

Familial Hypercholesterolemia (FH) is a monogenic disorder of lipid metabolism, resulting in severe dyslipidemia and increased cardiovascular disease risk. Simon Broome (SB) criteria for the diagnostic of FH are among the most frequently used in clinical setting, and are based on elevated total cholesterol (TC) and low density lipoprotein (LDLc) cholesterol levels, presence of tendinous xanthomas and family history, although only genetic testing can confirm the diagnosis. According to the Portuguese Study of Familial Hypercholesterolemia (EPHF), only around 42% of the patients with clinical criteria reveal a positive diagnostic for FH, a high false positive rate that represents a heavy burden in terms of healthcare costs. The main purpose of this work was to develop alternative classification models for FH diagnosis based on two different methods, logistic regression (LR) and decision trees (DT), using several biochemical indicators as predictor variables. Both models were compared with SB clinical criteria in terms of accuracy and efficiency, through bootstrap resampling techniques.