{"id":{"repo_id":"bu","oai_identifier":"oai:open.bu.edu:2144/53276"},"canonical_url":"https://search.ndltd.org/etd/bu/oai:open.bu.edu:2144/53276","repository":{"repo_id":"bu","name":"Boston University","base_url":"https://open.bu.edu/oai/request"},"display":{"title":"The Learn-As-you-GO (LAGO) design for continuous outcomes, a power goal to prevent failed trials, and implementation in R","abstract":"The Learn-As-you-GO (LAGO) design is an adaptive trial design that allows investigators to systematically optimize interventions based on accumulating data during an ongoing trial to prevent trial failure while maintaining the integrity of the trial. LAGO trials are typically multi-stage, multi-component intervention studies. After each stage, the results collected up to that stage are analyzed, the intervention package is reassessed, and a revised intervention package is implemented in the subsequent stage. By iteratively optimizing the intervention package (LAGO optimization), the LAGO design can identify the intervention package composition that minimizes the cost while ensuring that the expected outcome remains above a threshold (outcome goal). This dissertation addresses critical gaps in the LAGO methodology through three interconnected projects. First, we extend the LAGO design to accommodate trials with continuous outcomes using a Generalized Linear Model (GLM) framework. While previous work established LAGO for binary outcomes with logistic regression models, this extension significantly broadens its applicability to various areas of implementation trials where outcomes are measured on continuous scales. We demonstrate that the estimator of treatment effects from the final analysis based on all data is consistent and asymptotically normal, despite the dependencies introduced by learning between stages. Second, we incorporate statistical power directly into the LAGO optimizations. We introduce both unconditional and conditional power approaches, and we describe the calculation of the intervention component composition in the next stage while incorporating statistical power in LAGO optimizations. This project describes how the power goal affects the calculation of the intervention component compositions across various standard statistical tests. Last, we describe the development of a user-friendly R package for implementing the LAGO design. This R package bridges the gap between theoretical advances and practical applications by providing standardized functions for LAGO optimizations, including fitting outcome models and calculating the intervention component composition in the next stage. We illustrate the R package's functionality using built-in R datasets and real-world data from the BetterBirth study, a large-scale implementation trial that took place in Uttar Pradesh, India.","abstract_html":"The Learn-As-you-GO (LAGO) design is an adaptive trial design that allows investigators to systematically optimize interventions based on accumulating data during an ongoing trial to prevent trial failure while maintaining the integrity of the trial. LAGO trials are typically multi-stage, multi-component intervention studies. After each stage, the results collected up to that stage are analyzed, the intervention package is reassessed, and a revised intervention package is implemented in the subsequent stage. By iteratively optimizing the intervention package (LAGO optimization), the LAGO design can identify the intervention package composition that minimizes the cost while ensuring that the expected outcome remains above a threshold (outcome goal). This dissertation addresses critical gaps in the LAGO methodology through three interconnected projects. First, we extend the LAGO design to accommodate trials with continuous outcomes using a Generalized Linear Model (GLM) framework. While previous work established LAGO for binary outcomes with logistic regression models, this extension significantly broadens its applicability to various areas of implementation trials where outcomes are measured on continuous scales. We demonstrate that the estimator of treatment effects from the final analysis based on all data is consistent and asymptotically normal, despite the dependencies introduced by learning between stages. Second, we incorporate statistical power directly into the LAGO optimizations. We introduce both unconditional and conditional power approaches, and we describe the calculation of the intervention component composition in the next stage while incorporating statistical power in LAGO optimizations. This project describes how the power goal affects the calculation of the intervention component compositions across various standard statistical tests. Last, we describe the development of a user-friendly R package for implementing the LAGO design. This R package bridges the gap between theoretical advances and practical applications by providing standardized functions for LAGO optimizations, including fitting outcome models and calculating the intervention component composition in the next stage. We illustrate the R package&#x27;s functionality using built-in R datasets and real-world data from the BetterBirth study, a large-scale implementation trial that took place in Uttar Pradesh, India.","abstract_has_math":false,"creators":["Bing, Ante"],"institution":"Boston University","degree_name":"Doctor of Philosophy","degree_level":"doctoral","degree_discipline":"Mathematics & Statistics","degree_department":null,"school":null,"contributors":[],"advisors":["Lok, Judith J."],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025","date_published":"2025","updated_at":"2026-11-18T20:20:00Z","subjects":["Statistics","Adaptive clinical trial","Dependent sample","Implementation trial","Large-scale intervention trial","Public health"],"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":["0009-0001-4230-1675"],"render_values":[{"text":"0009-0001-4230-1675","href":"https://orcid.org/0009-0001-4230-1675","code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/2144/53276","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/2144/53276","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%2F53276","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Lok, Judith J."]},{"key":"dc:creator","label":"Author","values":["Bing, Ante"]},{"key":"dc:identifier.orcid","label":"ORCID","values":["0009-0001-4230-1675"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-07-09T19:12:54Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2026-07-09T19:12:54Z"]},{"key":"dc:date.issued","label":"Date","values":["2025"]},{"key":"dc:date.updated","label":"Dc Date Updated","values":["2026-07-09T16:15:25Z"]},{"key":"dc:type","label":"Dc Type","values":["Thesis/Dissertation"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Mathematics & Statistics"]},{"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":["Statistics","Adaptive clinical trial","Dependent sample","Implementation trial","Large-scale intervention trial","Public health"]}]},{"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/2144/53276"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["2025"]},{"key":"dc:description.abstract","label":"Abstract","values":["The Learn-As-you-GO (LAGO) design is an adaptive trial design that allows investigators to systematically optimize interventions based on accumulating data during an ongoing trial to prevent trial failure while maintaining the integrity of the trial. LAGO trials are typically multi-stage, multi-component intervention studies. After each stage, the results collected up to that stage are analyzed, the intervention package is reassessed, and a revised intervention package is implemented in the subsequent stage. By iteratively optimizing the intervention package (LAGO optimization), the LAGO design can identify the intervention package composition that minimizes the cost while ensuring that the expected outcome remains above a threshold (outcome goal). This dissertation addresses critical gaps in the LAGO methodology through three interconnected projects. First, we extend the LAGO design to accommodate trials with continuous outcomes using a Generalized Linear Model (GLM) framework. While previous work established LAGO for binary outcomes with logistic regression models, this extension significantly broadens its applicability to various areas of implementation trials where outcomes are measured on continuous scales. We demonstrate that the estimator of treatment effects from the final analysis based on all data is consistent and asymptotically normal, despite the dependencies introduced by learning between stages. Second, we incorporate statistical power directly into the LAGO optimizations. We introduce both unconditional and conditional power approaches, and we describe the calculation of the intervention component composition in the next stage while incorporating statistical power in LAGO optimizations. This project describes how the power goal affects the calculation of the intervention component compositions across various standard statistical tests. Last, we describe the development of a user-friendly R package for implementing the LAGO design. This R package bridges the gap between theoretical advances and practical applications by providing standardized functions for LAGO optimizations, including fitting outcome models and calculating the intervention component composition in the next stage. We illustrate the R package's functionality using built-in R datasets and real-world data from the BetterBirth study, a large-scale implementation trial that took place in Uttar Pradesh, India."]},{"key":"dc:title","label":"Title","values":["The Learn-As-you-GO (LAGO) design for continuous outcomes, a power goal to prevent failed trials, and implementation in R"]}]}],"canonical_facts":{"dc:contributor.advisor":["Lok, Judith J."],"dc:creator":["Bing, Ante"],"dc:date.accessioned":["2026-07-09T19:12:54Z"],"dc:date.available":["2026-07-09T19:12:54Z"],"dc:date.issued":["2025"],"dc:date.updated":["2026-07-09T16:15:25Z"],"dc:description":["2025"],"dc:description.abstract":["The Learn-As-you-GO (LAGO) design is an adaptive trial design that allows investigators to systematically optimize interventions based on accumulating data during an ongoing trial to prevent trial failure while maintaining the integrity of the trial. LAGO trials are typically multi-stage, multi-component intervention studies. After each stage, the results collected up to that stage are analyzed, the intervention package is reassessed, and a revised intervention package is implemented in the subsequent stage. By iteratively optimizing the intervention package (LAGO optimization), the LAGO design can identify the intervention package composition that minimizes the cost while ensuring that the expected outcome remains above a threshold (outcome goal). This dissertation addresses critical gaps in the LAGO methodology through three interconnected projects. First, we extend the LAGO design to accommodate trials with continuous outcomes using a Generalized Linear Model (GLM) framework. While previous work established LAGO for binary outcomes with logistic regression models, this extension significantly broadens its applicability to various areas of implementation trials where outcomes are measured on continuous scales. We demonstrate that the estimator of treatment effects from the final analysis based on all data is consistent and asymptotically normal, despite the dependencies introduced by learning between stages. Second, we incorporate statistical power directly into the LAGO optimizations. We introduce both unconditional and conditional power approaches, and we describe the calculation of the intervention component composition in the next stage while incorporating statistical power in LAGO optimizations. This project describes how the power goal affects the calculation of the intervention component compositions across various standard statistical tests. Last, we describe the development of a user-friendly R package for implementing the LAGO design. This R package bridges the gap between theoretical advances and practical applications by providing standardized functions for LAGO optimizations, including fitting outcome models and calculating the intervention component composition in the next stage. We illustrate the R package's functionality using built-in R datasets and real-world data from the BetterBirth study, a large-scale implementation trial that took place in Uttar Pradesh, India."],"dc:identifier.orcid":["0009-0001-4230-1675"],"dc:identifier.uri":["https://hdl.handle.net/2144/53276"],"dc:language.iso":["en_US"],"dc:rights":["Attribution 4.0 International"],"dc:rights.uri":["http://creativecommons.org/licenses/by/4.0/"],"dc:subject":["Statistics","Adaptive clinical trial","Dependent sample","Implementation trial","Large-scale intervention trial","Public health"],"dc:title":["The Learn-As-you-GO (LAGO) design for continuous outcomes, a power goal to prevent failed trials, and implementation in R"],"dc:type":["Thesis/Dissertation"],"thesis:degree_discipline":["Mathematics & Statistics"],"thesis:degree_level":["doctoral"],"thesis:degree_name":["Doctor of Philosophy"],"thesis:institution_name":["Boston University"]},"updated_at":"2026-11-18T20:20:00Z"}