Precision of Cancer Incidence Predictions Based on Poisson Distributed Observations

Show full item record

Title: Precision of Cancer Incidence Predictions Based on Poisson Distributed Observations
Author: Dyba, Tadeusz
Contributor organization: University of Helsinki, Faculty of Social Sciences, Department of Mathematics and Statistics
Helsingin yliopisto, Valtiotieteellinen tiedekunta, Matematiikan ja tilastotieteen laitos
Helsingfors universitet, Statsvetenskapliga fakulteten, Matematiska och statistiska institutionen
Date: 2000-06-30
Language: eng
Thesis level: Doctoral dissertation
Abstract: Predictions of disease incidence, including cancer incidence, play an important role in an epidemiological analysis, both in an administrative and scientific context. The use of reliable predictions in administrative planning guarantees optimal distribution of resources to diagnostics, treatment and rehabilitation. The prediction should always be accompanied by a prediction interval, a measure of its precision, in order to be properly utilized in a decisive process. In this study, a method is proposed for calculating approximate confidence and prediction intervals both for the numbers of cases and for the age-adjusted incidence rates by assuming Poisson or extra-Poisson distribution of stratum-specific number of incident cases, the assumption commonly accepted and used in epidemiology. The method can be applied to predict incidence cases based on a model with any functional form of linear predictor or link function. A good prediction model, taking into account the availability of reliable data, should produce a prediction close to the future true observation with as high precision as possible. Therefore, a group of simple linear and multiplicative age-period and age-period-region models, which are linear in time, has been also proposed. The models seem to represent well the main features of historical data and most of them preserve the age-specific incidence patterns of historical data in the period of prediction. Thanks to their simplicity, the precision of the predictions produced by them is also satisfactory. The investigation based on simulations confirmed a good reliability of the proposed approach to prediction, also when compared with alternative ones. The results of the simulations for models chosen showed good theoretical properties of the prediction interval, as based on the models, in terms of efficiency and accuracy. This property did not depend on the length of the calendar time intervals used for the estimation of the models. The successful application of the proposed approach for most cancer sites in Finland showed that it can be used quite generally and the software created can be easily applied for other data sets.
Description: Only abstract. Paper copies of master’s theses are listed in the Helka database ( Electronic copies of master’s theses are either available as open access or only on thesis terminals in the Helsinki University Library.Vain tiivistelmä. Sidottujen gradujen saatavuuden voit tarkistaa Helka-tietokannasta ( Digitaaliset gradut voivat olla luettavissa avoimesti verkossa tai rajoitetusti kirjaston opinnäytekioskeilla.Endast sammandrag. Inbundna avhandlingar kan sökas i Helka-databasen ( Elektroniska kopior av avhandlingar finns antingen öppet på nätet eller endast tillgängliga i bibliotekets avhandlingsterminaler.
Subject: incidence prediction - cancer
prediction interval
Poisson distribution
linear models
non-linear models
syöpä - ennusteet - matemaattiset mallit
Poissonin jakauma - sovellukset

Files in this item

Total number of downloads: Loading...

Files Size Format View
abstract.pdf 48.81Kb PDF View/Open

This item appears in the following Collection(s)

Show full item record