Publication Results



Author: Ieva, Francesca
Title: Modelli statistici per lo studio dei tempi di intervento nell infarto miocardico acuto
Date: Thursday 23rd October 2008
Advisor: Paganoni, A.m.
Advisor II: Sesana, G.
Download link:
Abstract: This work arises from a collaboration with the Working Group for Cardiac Emergency (ACEU) of Lombardia Region, the Dispatch Center of 118 (the national free number for medical emergencies) and the Hospital Niguarda Ca' Granda, in order to create and organize the Cardiological Network between the hospitals in the Lombard area. This collaboration has given birth to a Strategic Project focused on comprehension, monitoring and organization of the terapies related to the Acute Coronary Sindromes (ACS) in Lombardia. Up to now, a study called MOMI2, consisting in data collections during five survey periods, has been conducted on ST-segment elevation Myocardial Infarction (STEMI). The topic of my master thesis is concerned with the clinical request of analysis of data coming from dimission case sheets of patients with STEMI diagnosis admitted to hospitals which belong to the Network. The aim of this is to give a preliminar description of these data and identify which covariates represent significant prognostic factors. Furthermore we studied advanced statistical methods, which probably will be the natural continuation of the studies conducted untill now. Infact statistical analises of these data consitute valid supports to important medical decisions. The structure of the work is as follows: in Chapter 1 we present the problem, a description of the disease (STEMI) and the dataset; in Chapter 2 we present the theory about Contingency Tables and their association parameters (such as Odds Ratios); in Chapter 3 we present the study of exact inferential techniques for building Confidence Interval for the previous parameters in the case of small samples, which is typical of our dataset; in Chapter 4 we present the theory of Generalized Linear Model (GLM), with particular attention to Logistic Regression Models. We also study the exact inference that can be conducted on these models using Monte Carlo sampling techniques; in Chapter 5 we present the theory of Generalized Additive Models (GAM), a non parametric technique which generalises the one presented in the previous Chapter; finally, in Chapter 6 we use all these tools to make statistical analises on data coming from MOMI2 dataset. All the analises have been carried out using the statistical software R.2.7.0. The fundamental result of this thesis is not only the study of advanced and innovative statistical techniques, but also the social impact of the results themselves, obtained thanks to the synergic interaction between statisticians and physicians.