In this thesis, I apply three of the existing meta analysis approaches on the receiver operating characteristic curves(ROC Curves). The purpose is to find out an efficient way to synthesize the ROC curves from multiple individual studies. By transforming the ROC Curves into linear models, I apply three meta analysis methods to synthesize their coefficients. The three methods deal with within and between studies correlation in three different ways. To decide which method gives the arguably best result, we used an extra set of data as a test. And I use the prediction error as a criterion of the three synthesized ROC Curves.

Creator

• Yang, Shuang

Contributors

• Peiris, Thelge B.

Degree

• MS

Unit

• Mathematical Sciences

Publisher

• Worcester Polytechnic Institute

Identifier

• etd-23211

Advisor

• Peiris, Buddika

Defense date

• 2021-05-06

Year

• 2021

Date created

• 2021-05-06

Resource type

• Thesis

Rights statement

• In Copyright

Last modified

• 2021-08-29