{"id":{"repo_id":"bu","oai_identifier":"oai:open.bu.edu:2144/53618"},"canonical_url":"https://search.ndltd.org/etd/bu/oai:open.bu.edu:2144/53618","repository":{"repo_id":"bu","name":"Boston University","base_url":"https://open.bu.edu/oai/request"},"display":{"title":"Methodological advancements for synthetic data generation","abstract":"Synthetic data generation is a promising solution to data access barriers in public health research by enabling researchers to share and analyze data while maintaining patient privacy. However, existing frameworks for synthetic data generation lack the distributional awareness, accessibility, and rigorous evaluation needed to generate high fidelity synthetic public health datasets. In this dissertation we will first introduce AutoTab as the first R package for tabular variational autoencoder (VAE) implementation. This package provides researchers with extensive customization options and a distributionally weighted evidence lower bound (wTAB-ELBO). In the second project, we present Distributionally Aware Tabular VAE (DA-TVAE), an extension of the AutoTab framework that introduces variable-type specific distributional choices, including extensions to count, ordinal, and non-Gaussian continuous variables. In the third project, we propose the Structures Laplacian Multi-distributional Maximum Mean Discrepancy (SLM MMD), a novel tabular fidelity metric. SLM MMD can be used for hyperparameter tuning, has attractive theoretical properties, and is less computationally expensive than existing methods. Together these three contributions form a complete pipeline for VAE-based synthetic data generation, tuning, and evaluation for tabular public health data.","abstract_html":"Synthetic data generation is a promising solution to data access barriers in public health research by enabling researchers to share and analyze data while maintaining patient privacy. However, existing frameworks for synthetic data generation lack the distributional awareness, accessibility, and rigorous evaluation needed to generate high fidelity synthetic public health datasets. In this dissertation we will first introduce AutoTab as the first R package for tabular variational autoencoder (VAE) implementation. This package provides researchers with extensive customization options and a distributionally weighted evidence lower bound (wTAB-ELBO). In the second project, we present Distributionally Aware Tabular VAE (DA-TVAE), an extension of the AutoTab framework that introduces variable-type specific distributional choices, including extensions to count, ordinal, and non-Gaussian continuous variables. In the third project, we propose the Structures Laplacian Multi-distributional Maximum Mean Discrepancy (SLM MMD), a novel tabular fidelity metric. SLM MMD can be used for hyperparameter tuning, has attractive theoretical properties, and is less computationally expensive than existing methods. Together these three contributions form a complete pipeline for VAE-based synthetic data generation, tuning, and evaluation for tabular public health data.","abstract_has_math":false,"creators":["Milligan, Sarah"],"institution":"Boston University","degree_name":"Doctor of Philosophy","degree_level":"doctoral","degree_discipline":"Biostatistics","degree_department":null,"school":null,"contributors":[],"advisors":["Weinberg, Janice","Shafie Khorassani, Fatema"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T19:24:51Z","subjects":["Biostatistics","Data sharing","Fidelity","Generative modeling","Synthetic data","Variational autoencoder"],"languages":["en_US"],"rights":["Attribution 4.0 International"],"rights_urls":["http://creativecommons.org/licenses/by/4.0/"],"identifier_entries":[],"person_identifier_entries":[{"key":"dc:identifier.orcid","label":"ORCID","values":["0009-0009-2860-618X"],"render_values":[{"text":"0009-0009-2860-618X","href":"https://orcid.org/0009-0009-2860-618X","code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/2144/53618","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/2144/53618","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://open.bu.edu/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Aopen.bu.edu%3A2144%2F53618","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Weinberg, Janice","Shafie Khorassani, Fatema"]},{"key":"dc:creator","label":"Author","values":["Milligan, Sarah"]},{"key":"dc:identifier.orcid","label":"ORCID","values":["0009-0009-2860-618X"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-10-05T19:06:58Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2026-10-05T19:06:58Z"]},{"key":"dc:date.issued","label":"Date","values":["2026"]},{"key":"dc:date.updated","label":"Dc Date Updated","values":["2026-10-02T16:15:18Z"]},{"key":"dc:type","label":"Dc Type","values":["Thesis/Dissertation"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Biostatistics"]},{"key":"thesis:degree_level","label":"Degree Level","values":["doctoral"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["Boston University"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Biostatistics","Data sharing","Fidelity","Generative modeling","Synthetic data","Variational autoencoder"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en_US"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution 4.0 International"]},{"key":"dc:rights.uri","label":"Rights URI","values":["http://creativecommons.org/licenses/by/4.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/2144/53618"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["2026"]},{"key":"dc:description.abstract","label":"Abstract","values":["Synthetic data generation is a promising solution to data access barriers in public health research by enabling researchers to share and analyze data while maintaining patient privacy. However, existing frameworks for synthetic data generation lack the distributional awareness, accessibility, and rigorous evaluation needed to generate high fidelity synthetic public health datasets. In this dissertation we will first introduce AutoTab as the first R package for tabular variational autoencoder (VAE) implementation. This package provides researchers with extensive customization options and a distributionally weighted evidence lower bound (wTAB-ELBO). In the second project, we present Distributionally Aware Tabular VAE (DA-TVAE), an extension of the AutoTab framework that introduces variable-type specific distributional choices, including extensions to count, ordinal, and non-Gaussian continuous variables. In the third project, we propose the Structures Laplacian Multi-distributional Maximum Mean Discrepancy (SLM MMD), a novel tabular fidelity metric. SLM MMD can be used for hyperparameter tuning, has attractive theoretical properties, and is less computationally expensive than existing methods. Together these three contributions form a complete pipeline for VAE-based synthetic data generation, tuning, and evaluation for tabular public health data."]},{"key":"dc:title","label":"Title","values":["Methodological advancements for synthetic data generation"]}]}],"canonical_facts":{"dc:contributor.advisor":["Weinberg, Janice","Shafie Khorassani, Fatema"],"dc:creator":["Milligan, Sarah"],"dc:date.accessioned":["2026-10-05T19:06:58Z"],"dc:date.available":["2026-10-05T19:06:58Z"],"dc:date.issued":["2026"],"dc:date.updated":["2026-10-02T16:15:18Z"],"dc:description":["2026"],"dc:description.abstract":["Synthetic data generation is a promising solution to data access barriers in public health research by enabling researchers to share and analyze data while maintaining patient privacy. However, existing frameworks for synthetic data generation lack the distributional awareness, accessibility, and rigorous evaluation needed to generate high fidelity synthetic public health datasets. In this dissertation we will first introduce AutoTab as the first R package for tabular variational autoencoder (VAE) implementation. This package provides researchers with extensive customization options and a distributionally weighted evidence lower bound (wTAB-ELBO). In the second project, we present Distributionally Aware Tabular VAE (DA-TVAE), an extension of the AutoTab framework that introduces variable-type specific distributional choices, including extensions to count, ordinal, and non-Gaussian continuous variables. In the third project, we propose the Structures Laplacian Multi-distributional Maximum Mean Discrepancy (SLM MMD), a novel tabular fidelity metric. SLM MMD can be used for hyperparameter tuning, has attractive theoretical properties, and is less computationally expensive than existing methods. Together these three contributions form a complete pipeline for VAE-based synthetic data generation, tuning, and evaluation for tabular public health data."],"dc:identifier.orcid":["0009-0009-2860-618X"],"dc:identifier.uri":["https://hdl.handle.net/2144/53618"],"dc:language.iso":["en_US"],"dc:rights":["Attribution 4.0 International"],"dc:rights.uri":["http://creativecommons.org/licenses/by/4.0/"],"dc:subject":["Biostatistics","Data sharing","Fidelity","Generative modeling","Synthetic data","Variational autoencoder"],"dc:title":["Methodological advancements for synthetic data generation"],"dc:type":["Thesis/Dissertation"],"thesis:degree_discipline":["Biostatistics"],"thesis:degree_level":["doctoral"],"thesis:degree_name":["Doctor of Philosophy"],"thesis:institution_name":["Boston University"]},"updated_at":"2026-11-18T19:24:51Z"}