Tıbbi laboratuvar testlerinin pediyatrik referans aralıklarının belirlenmesi için bazı veri madenciliği teknikleri ile deneyler

Aim of this thesis is to conduct some experiments on different datasets by using some data mining techniques under the multi-disciplinary study of establishing pediatric reference intervals from laboratory test results. While traditional methods can be applied for adults, it is difficult and inconvenient to apply these rules for pediatric patient groups. In order to resolve this need, this study has been performed to make the reference interval establishment process easier with data mining techniques. To achieve this goal, a tool has been developed to allow specialists to easily load laboratory test data and to perform rapid data mining operations. With the help of this tool, specialists will be able to apply data mining techniques such as filtering, examine the distribution of parameters age-, gender-, and diagnosis-related data and apply machine learning algorithms in their reference interval studies based on hospital or device and publish it after reviewing their clinical accuracy. Pediatric reference intervals established by The Canadian Laboratory Initiative on Pediatric Reference Intervals (CALIPER) are used for the experiments on developed tool and the results were compared with the reference intervals published by CALIPER. As a result, by using data mining techniques via developed tool instead of classical statistical methods through the process of establishing reference intervals from laboratory test data, we can obtain faster and more effective results.