{"id":{"repo_id":"ku","oai_identifier":"oai:kuscholarworks.ku.edu:1808/36417"},"canonical_url":"https://search.dev.ndltd.org/etd/ku/oai:kuscholarworks.ku.edu:1808/36417","repository":{"repo_id":"ku","name":"University of Kansas","base_url":"https://kuscholarworks.ku.edu/server/oai/request"},"display":{"title":"Advances in estimation and inference around a biomarker’s true and false classification rates under trichotomous settings","abstract":"Discovering and evaluating new biomarkers for the detection of a disease is a topic of globalinterest, especially when dealing with cancer settings. As with many other diseases, cancer has a progressive nature. Therefore, it is not uncommon that in biomarker studies, investigators recruit subjects that fall into one of three categories: healthy, benign, and aggressive. In these configurations (trichotomous setting), the usual ROC analysis needs to be considered in a 3-dimensional framework. In this thesis, under the framework of the 3-dimensional ROC analysis, I propose new parametric, semi-parametric, and non-parametric approaches that can improve the assessment of biomarkers. This is done by simultaneously taking all three groups into consideration in the inferential process of the optimal cutoffs. In this thesis, we also explore incorporating the Multiple Imputation technique to improve the ROC estimation in biomarker studies that deal with missing data. Finally, in our last chapter, we present a user-friendly software for the derivation of biomarker cutoffs in trichotomous settings that can accommodate misclassification costs. We apply our approaches to a real data set that involves hepatocellular carcinoma (HCC)patients. HCC is the most common cancer of the liver and the ninth leading cause of cancerrelated death in the United States. Our approaches can provide deeper insights regarding the accuracy of the candidate biomarker, improved cutoff estimation, as well as biomarker evaluation in the context of missing data.","abstract_html":"Discovering and evaluating new biomarkers for the detection of a disease is a topic of globalinterest, especially when dealing with cancer settings. As with many other diseases, cancer has a progressive nature. Therefore, it is not uncommon that in biomarker studies, investigators recruit subjects that fall into one of three categories: healthy, benign, and aggressive. In these configurations (trichotomous setting), the usual ROC analysis needs to be considered in a 3-dimensional framework. In this thesis, under the framework of the 3-dimensional ROC analysis, I propose new parametric, semi-parametric, and non-parametric approaches that can improve the assessment of biomarkers. This is done by simultaneously taking all three groups into consideration in the inferential process of the optimal cutoffs. In this thesis, we also explore incorporating the Multiple Imputation technique to improve the ROC estimation in biomarker studies that deal with missing data. Finally, in our last chapter, we present a user-friendly software for the derivation of biomarker cutoffs in trichotomous settings that can accommodate misclassification costs. We apply our approaches to a real data set that involves hepatocellular carcinoma (HCC)patients. HCC is the most common cancer of the liver and the ninth leading cause of cancerrelated death in the United States. Our approaches can provide deeper insights regarding the accuracy of the candidate biomarker, improved cutoff estimation, as well as biomarker evaluation in the context of missing data.","abstract_has_math":false,"creators":["Shi, Peng"],"institution":"University of Kansas","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Bantis, Leonidas"],"committee_chairs":[],"committee_members":[],"year":2022,"date_issued":"2022-12-31","date_published":"2022-12-31","updated_at":"2026-07-24T02:45:19Z","subjects":["Biostatistics","cutoff","Euclidean distance","misclassification","Multiple imputation","ROC","true class fraction"],"languages":["en"],"rights":["Copyright held by the author."],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["http://dissertations.umi.com/ku:18678"],"render_values":[{"text":"http://dissertations.umi.com/ku:18678","href":"http://dissertations.umi.com/ku:18678","code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/1808/36417","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Bantis, Leonidas"]},{"key":"dc:creator","label":"Author","values":["Shi, Peng"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-03-10T19:33:19Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2026-03-10T19:33:19Z"]},{"key":"dc:date.issued","label":"Date","values":["2022-12-31"]},{"key":"dc:publisher","label":"Institution","values":["University of Kansas"]},{"key":"dc:type","label":"Dc Type","values":["Dissertation"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Biostatistics","cutoff","Euclidean distance","misclassification","Multiple imputation","ROC","true class fraction"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright held by the author."]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["http://dissertations.umi.com/ku:18678"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/1808/36417"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Discovering and evaluating new biomarkers for the detection of a disease is a topic of globalinterest, especially when dealing with cancer settings. As with many other diseases, cancer has a progressive nature. Therefore, it is not uncommon that in biomarker studies, investigators recruit subjects that fall into one of three categories: healthy, benign, and aggressive. In these configurations (trichotomous setting), the usual ROC analysis needs to be considered in a 3-dimensional framework. In this thesis, under the framework of the 3-dimensional ROC analysis, I propose new parametric, semi-parametric, and non-parametric approaches that can improve the assessment of biomarkers. This is done by simultaneously taking all three groups into consideration in the inferential process of the optimal cutoffs. In this thesis, we also explore incorporating the Multiple Imputation technique to improve the ROC estimation in biomarker studies that deal with missing data. Finally, in our last chapter, we present a user-friendly software for the derivation of biomarker cutoffs in trichotomous settings that can accommodate misclassification costs. We apply our approaches to a real data set that involves hepatocellular carcinoma (HCC)patients. HCC is the most common cancer of the liver and the ninth leading cause of cancerrelated death in the United States. Our approaches can provide deeper insights regarding the accuracy of the candidate biomarker, improved cutoff estimation, as well as biomarker evaluation in the context of missing data."]},{"key":"dc:title","label":"Title","values":["Advances in estimation and inference around a biomarker’s true and false classification rates under trichotomous settings"]}]}],"canonical_facts":{"dc:contributor.advisor":["Bantis, Leonidas"],"dc:creator":["Shi, Peng"],"dc:date.accessioned":["2026-03-10T19:33:19Z"],"dc:date.available":["2026-03-10T19:33:19Z"],"dc:date.issued":["2022-12-31"],"dc:description.abstract":["Discovering and evaluating new biomarkers for the detection of a disease is a topic of globalinterest, especially when dealing with cancer settings. As with many other diseases, cancer has a progressive nature. Therefore, it is not uncommon that in biomarker studies, investigators recruit subjects that fall into one of three categories: healthy, benign, and aggressive. In these configurations (trichotomous setting), the usual ROC analysis needs to be considered in a 3-dimensional framework. In this thesis, under the framework of the 3-dimensional ROC analysis, I propose new parametric, semi-parametric, and non-parametric approaches that can improve the assessment of biomarkers. This is done by simultaneously taking all three groups into consideration in the inferential process of the optimal cutoffs. In this thesis, we also explore incorporating the Multiple Imputation technique to improve the ROC estimation in biomarker studies that deal with missing data. Finally, in our last chapter, we present a user-friendly software for the derivation of biomarker cutoffs in trichotomous settings that can accommodate misclassification costs. We apply our approaches to a real data set that involves hepatocellular carcinoma (HCC)patients. HCC is the most common cancer of the liver and the ninth leading cause of cancerrelated death in the United States. Our approaches can provide deeper insights regarding the accuracy of the candidate biomarker, improved cutoff estimation, as well as biomarker evaluation in the context of missing data."],"dc:identifier.other":["http://dissertations.umi.com/ku:18678"],"dc:identifier.uri":["https://hdl.handle.net/1808/36417"],"dc:language.iso":["en"],"dc:publisher":["University of Kansas"],"dc:rights":["Copyright held by the author."],"dc:subject":["Biostatistics","cutoff","Euclidean distance","misclassification","Multiple imputation","ROC","true class fraction"],"dc:title":["Advances in estimation and inference around a biomarker’s true and false classification rates under trichotomous settings"],"dc:type":["Dissertation"]},"updated_at":"2026-07-24T02:45:19Z"}