Analiza policističnog sindroma jajnika logističkom regresijom i diskriminantnom analizom

master's thesis

• Title (english): Analysis of polycystic ovary syndrome by logistic regression and discriminant analysis
• Author: Teodora Peček
• Mentor: Anamarija Jazbec (mentor)
• Committee member: Anamarija Jazbec (predsjednik povjerenstva)
• Committee member: Marko Vrdoljak (član povjerenstva)
• Committee member: Nenad Antonić (član povjerenstva)
• Committee member: Nina Kamčev (član povjerenstva)
• Granter: University of Zagreb; Faculty of Science; (Department of Mathematics); Zagreb
• Defense date and country: 2024-02-28, Croatia
• Scientific / art field, discipline and subdiscipline: NATURAL SCIENCES; Mathematics

Abstract

U ovom radu uvedeni su osnovni pojmovi logističke regresije - logistički model, izgled, omjer izgleda. Opisano je kako se testira adekvatnost logističkog modela, interpretiraju dobiveni parametri i biraju nezavisne varijable modela. Uveden je pojam diskriminantne analize i poseban slučaj linearne diskriminantne analize. Opisano je kako se provodi validacija dobivenih rezultata i kako se biraju prediktorske varijable. Korištenjem programa SAS OnDemand analizirana je baza podataka pacijentica iz nekoliko bolnica u Indiji kako bi se odredilo koje varijable utječu na pojavnost sindroma policističnih jajnika. Dobiveni su univarijatni logistički modeli, multivarijatni logistički model i logistički model stepwise procedure. Varijable koje su značajne u svim modelima su Follicle No L (broj folikularnih cista na lijevom jajniku), Follicle No R (broj folikularnih cista na desnom jajniku), Cycle (redovitost menstrualnog ciklusa), Weight gain (povećanje tjelesne težine), Hair growth (pojačan rasta dlaka), Skin darkening (hiperpigmentacija kože) i Acne (akne). Korištena je direktna linearna diskriminantna analiza i stepwise procedura linearne diskriminantne analize za klasifikaciju podataka. Stepwise procedurom diskriminantne analize izabrane su varijable Follicle No R, Weight gain, Hair growth, Cycle, Skin darkening, AMH (razina anti-Mullerovog hormona), Acne, Age (starost), RBS (razina glukoze u krvi), Follicle No L

Abstract (english)

In this thesis basic terms of logistic regression are introduced - logistic model, odds, odds ratio. It is described how to test the adequacy of the logistic model, interpret the obtained parameters and select the independent variables of the model. The concept of discriminant analysis and the special case of linear discriminant analysis are introduced. It is described how the validation of the obtained results is carried out and how the predictor variables are selected. Using SAS OnDemand, a database of patients from several hospitals in India was analyzed to determine which variables influence the incidence of polycystic ovary syndrome. Univariate logistic models, multivariate logistic model and logistic model using stepwise procedure were obtained. Variables that are significant in all models are Follicle No L (number of follicular cysts on the left ovary), Follicle No R (number of follicular cysts on the right ovary), Cycle (regularity of the menstrual cycle), Weight gain , Hair growth, Skin darkening and Acne. Direct linear discriminant analysis and linear discriminant analysis using stepwise procedure were used for data classification. Stepwise discriminant analysis selected the variables Follicle No R, Weight gain, Hair growth, Cycle, Skin darkening, AMH (anti-Mullerian hormone level), Acne, Age, RBS (glucose level in blood), Follicle No L.

• Keywords: logistički model; logistički izgled; omjer izgleda; SAS OnDemand
• Language: croatian
• URN:NBN: urn:nbn:hr:217:709033
• Study programme: Title: Mathematical Statistics; Study programme type: university; Study level: graduate; Academic / professional title: sveučilišni magistar matematike (sveučilišni magistar matematike)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-03-14 11:29:01