{"id":{"repo_id":"mit","oai_identifier":"oai:dspace.mit.edu:1721.1/139482"},"canonical_url":"https://search.dev.ndltd.org/etd/mit/oai:dspace.mit.edu:1721.1/139482","repository":{"repo_id":"mit","name":"MIT","base_url":"https://dspace.mit.edu/oai/request"},"display":{"title":"Novel Machine Learning Algorithms for Personalized Medicine and Insurance","abstract":"Over the past decades, analytics have provided the promise of revolutionizing healthcare, providing more effective, patient-centered, and personalized care. As an increasing amount of data is being collected, computational performance is improved, and new algorithms are developed, machine learning has been viewed as the key analytical tool that will advance healthcare delivery. Nevertheless, until recently, despite the enthusiasm about the potential of “big data”, only a few examples have impacted the current clinical practice. This thesis presents a combination of predictive and prescriptive methodologies that will empower the transition to personalized medicine. We propose new machine learning algorithms to address major data imperfections like missing values, censored observations, and unobserved counterfactuals. Leveraging a wide variety of data sources, including health and claims records, longitudinal studies, and unstructured medical reports, we demonstrate the potential benefit of analytics in the context of cardiovascular and cerebrovascular diseases. To propel the adoption of these methodologies, we lay the foundations in the area of algorithmic insurance, proposing a quantitative framework to estimate the litigation risk of machine learning models. This work emphasizes interpretability and the design of models that facilitate clinician engagement and integration into the healthcare system. Part I introduces data-driven algorithms for missing data imputation, clustering, and survival analysis that lie at the intersection of machine learning and optimization. Part II highlights the potential of prescriptive and predictive analytics in the medical field. We develop a new framework for personalized prescriptions and apply it for the treatment of coronary artery disease. Part II also presents predictive models that could support the early diagnosis and improve the management of stroke patients. Finally, Part III proposes a novel risk evaluation methodology that will enable healthcare institutions to manage the risk exposure resulting from the implementation of analytical decision tools.","abstract_html":"Over the past decades, analytics have provided the promise of revolutionizing healthcare, providing more effective, patient-centered, and personalized care. As an increasing amount of data is being collected, computational performance is improved, and new algorithms are developed, machine learning has been viewed as the key analytical tool that will advance healthcare delivery. Nevertheless, until recently, despite the enthusiasm about the potential of “big data”, only a few examples have impacted the current clinical practice. This thesis presents a combination of predictive and prescriptive methodologies that will empower the transition to personalized medicine. We propose new machine learning algorithms to address major data imperfections like missing values, censored observations, and unobserved counterfactuals. Leveraging a wide variety of data sources, including health and claims records, longitudinal studies, and unstructured medical reports, we demonstrate the potential benefit of analytics in the context of cardiovascular and cerebrovascular diseases. To propel the adoption of these methodologies, we lay the foundations in the area of algorithmic insurance, proposing a quantitative framework to estimate the litigation risk of machine learning models. This work emphasizes interpretability and the design of models that facilitate clinician engagement and integration into the healthcare system. Part I introduces data-driven algorithms for missing data imputation, clustering, and survival analysis that lie at the intersection of machine learning and optimization. Part II highlights the potential of prescriptive and predictive analytics in the medical field. We develop a new framework for personalized prescriptions and apply it for the treatment of coronary artery disease. Part II also presents predictive models that could support the early diagnosis and improve the management of stroke patients. Finally, Part III proposes a novel risk evaluation methodology that will enable healthcare institutions to manage the risk exposure resulting from the implementation of analytical decision tools.","abstract_has_math":false,"creators":["Orfanoudaki, Agni"],"institution":"Massachusetts Institute of Technology","degree_name":"Doctoral","degree_level":null,"degree_discipline":null,"degree_department":"Massachusetts Institute of Technology. Operations Research Center","school":null,"contributors":[],"advisors":["Bertsimas, Dimitris"],"committee_chairs":[],"committee_members":[],"year":2021,"date_issued":"2021-06","date_published":"2021-06","updated_at":"2026-07-22T22:21:47Z","subjects":[],"languages":[],"rights":["In Copyright - Educational Use Permitted","Copyright MIT"],"rights_urls":["http://rightsstatements.org/page/InC-EDU/1.0/"],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/1721.1/139482","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Bertsimas, Dimitris"]},{"key":"dc:contributor.department","label":"Department","values":["Massachusetts Institute of Technology. Operations Research Center"]},{"key":"dc:creator","label":"Author","values":["Orfanoudaki, Agni"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2022-01-14T15:14:40Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2022-01-14T15:14:40Z"]},{"key":"dc:date.issued","label":"Date","values":["2021-06"]},{"key":"dc:publisher","label":"Institution","values":["Massachusetts Institute of Technology"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctoral","Doctor of Philosophy"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:rights","label":"Dc Rights","values":["In Copyright - Educational Use Permitted","Copyright MIT"]},{"key":"dc:rights.uri","label":"Rights URI","values":["http://rightsstatements.org/page/InC-EDU/1.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/1721.1/139482"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Over the past decades, analytics have provided the promise of revolutionizing healthcare, providing more effective, patient-centered, and personalized care. As an increasing amount of data is being collected, computational performance is improved, and new algorithms are developed, machine learning has been viewed as the key analytical tool that will advance healthcare delivery. Nevertheless, until recently, despite the enthusiasm about the potential of “big data”, only a few examples have impacted the current clinical practice. This thesis presents a combination of predictive and prescriptive methodologies that will empower the transition to personalized medicine. We propose new machine learning algorithms to address major data imperfections like missing values, censored observations, and unobserved counterfactuals. Leveraging a wide variety of data sources, including health and claims records, longitudinal studies, and unstructured medical reports, we demonstrate the potential benefit of analytics in the context of cardiovascular and cerebrovascular diseases. To propel the adoption of these methodologies, we lay the foundations in the area of algorithmic insurance, proposing a quantitative framework to estimate the litigation risk of machine learning models. This work emphasizes interpretability and the design of models that facilitate clinician engagement and integration into the healthcare system. Part I introduces data-driven algorithms for missing data imputation, clustering, and survival analysis that lie at the intersection of machine learning and optimization. Part II highlights the potential of prescriptive and predictive analytics in the medical field. We develop a new framework for personalized prescriptions and apply it for the treatment of coronary artery disease. Part II also presents predictive models that could support the early diagnosis and improve the management of stroke patients. Finally, Part III proposes a novel risk evaluation methodology that will enable healthcare institutions to manage the risk exposure resulting from the implementation of analytical decision tools."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Ph.D."]},{"key":"dc:title","label":"Title","values":["Novel Machine Learning Algorithms for Personalized Medicine and Insurance"]}]}],"canonical_facts":{"dc:contributor.advisor":["Bertsimas, Dimitris"],"dc:contributor.department":["Massachusetts Institute of Technology. Operations Research Center"],"dc:creator":["Orfanoudaki, Agni"],"dc:date.accessioned":["2022-01-14T15:14:40Z"],"dc:date.available":["2022-01-14T15:14:40Z"],"dc:date.issued":["2021-06"],"dc:description.abstract":["Over the past decades, analytics have provided the promise of revolutionizing healthcare, providing more effective, patient-centered, and personalized care. As an increasing amount of data is being collected, computational performance is improved, and new algorithms are developed, machine learning has been viewed as the key analytical tool that will advance healthcare delivery. Nevertheless, until recently, despite the enthusiasm about the potential of “big data”, only a few examples have impacted the current clinical practice. This thesis presents a combination of predictive and prescriptive methodologies that will empower the transition to personalized medicine. We propose new machine learning algorithms to address major data imperfections like missing values, censored observations, and unobserved counterfactuals. Leveraging a wide variety of data sources, including health and claims records, longitudinal studies, and unstructured medical reports, we demonstrate the potential benefit of analytics in the context of cardiovascular and cerebrovascular diseases. To propel the adoption of these methodologies, we lay the foundations in the area of algorithmic insurance, proposing a quantitative framework to estimate the litigation risk of machine learning models. This work emphasizes interpretability and the design of models that facilitate clinician engagement and integration into the healthcare system. Part I introduces data-driven algorithms for missing data imputation, clustering, and survival analysis that lie at the intersection of machine learning and optimization. Part II highlights the potential of prescriptive and predictive analytics in the medical field. We develop a new framework for personalized prescriptions and apply it for the treatment of coronary artery disease. Part II also presents predictive models that could support the early diagnosis and improve the management of stroke patients. Finally, Part III proposes a novel risk evaluation methodology that will enable healthcare institutions to manage the risk exposure resulting from the implementation of analytical decision tools."],"dc:description.degree":["Ph.D."],"dc:identifier.uri":["https://hdl.handle.net/1721.1/139482"],"dc:publisher":["Massachusetts Institute of Technology"],"dc:rights":["In Copyright - Educational Use Permitted","Copyright MIT"],"dc:rights.uri":["http://rightsstatements.org/page/InC-EDU/1.0/"],"dc:title":["Novel Machine Learning Algorithms for Personalized Medicine and Insurance"],"dc:type":["Thesis"],"thesis:degree_name":["Doctoral","Doctor of Philosophy"]},"updated_at":"2026-07-22T22:21:47Z"}