{"id":{"repo_id":"edinburgh","oai_identifier":"oai:era.ed.ac.uk:1842/43167"},"canonical_url":"https://search.ndltd.org/etd/edinburgh/oai:era.ed.ac.uk:1842/43167","repository":{"repo_id":"edinburgh","name":"University of Edinburgh","base_url":"https://era.ed.ac.uk/server/oai/request"},"display":{"title":"Supporting health data science education with learning analytics and artificial intelligence","abstract":"Health Data Science (HDS) is a new and interdisciplinary field that leverages computational methods, such as data science, statistics, and artificial intelligence, to analyse health, medical, or biological data. There is limited knowledge about how to teach and learn HDS topics effectively. Despite the urgent need to train health data scientists, educational institutions worldwide struggle to teach these interdisciplinary concepts effectively. This challenge is particularly evident in postgraduate studies, where learners come from diverse backgrounds. Additionally, there is a lack of clear guidelines on effective learning strategies in HDS to support students in addressing their educational challenges. To address this gap, this thesis aims to offer insights into health data science education through three key objectives. Firstly, it aims to uncover engagement patterns with HDS resources, including how learners interact with various types of educational resources and topics (Objective 1). Secondly, it aims to identify students’ learning tactics and strategies in health data science and explore how they impact student performance (Objective 2). Lastly, the thesis aims to develop an Artificial Intelligence (AI) assistant for the early prediction of struggling students in HDS and to provide them with personalised suggestions (Objective 3). A range of methods were employed, including qualitative content analysis, quantitative analysis of questionnaire data, and AI techniques on log data from online learning platforms. The qualitative content analysis by interviewing early career researchers from two UK institutes was conducted to understand and elaborate on the challenges they face in HDS education. Subsequent analysis then explored the self-reported learning strategies used by these researchers to overcome their educational challenges while shedding light on effective teaching methods and course design in HDS. A survey questionnaire was used to identify learning strategies and course design preferences in HDS among undergraduate and postgraduate learners across three HDS courses with varied modalities, designs, and academic levels. AI methods were used to uncover learners’ engagement with different resources as well as learning tactics and strategies based on clickstream data, reflecting the actual behaviours of learners. The thesis also analysed the correlation between the use of these learning tactics and strategies and students’ performance. Finally, a novel AI assistant was developed based on HDS students’ behaviours and learning process. The AI method used explainable AI and a counterfactual approach to provide suggestions that can help improve students’ performance. The results related to the first objective of this thesis indicate that students in HDS preferred a variety of resources and believed visually impressive materials enhance their understanding of HDS topics. Among these topics, machine learning emerged as the most interesting, while analysing genomic data was identified as the most difficult topic. Regarding the second objective, a range of learning tactics and strategies were identified, among which, connecting new information to prior knowledge and peer learning showed the highest correlation with student performance. The results also showed that students who engaged deeply with the material and applied different learning tactics throughout the course achieved higher grades than others. For the third objective, the novel AI assistant showed strong accuracy in predicting students’ performance, with an Area Under Curve (AUC) between 83% and 91%. The feedback generated by the AI, as evaluated by the course organiser, confirmed the validity of the suggestions. Additionally, the evaluation of the AI method on an external dataset outperformed traditional machine learning algorithms. The results of this thesis can potentially improve HDS education in the following ways. It can help develop effective courses through the appropriate design of learning materials (Objective 1). Also, evidence from Objective 2 can lead to guidelines that can help learners better manage their educational challenges. The AI model developed (Objective 3) can be used by teachers to identify students needing help and provide them with personalised suggestions according to their learning process. The evidence-based guidelines derived from this thesis have the potential to empower both learners and educators in HDS. We consider these to be significant contributions to the critical yet under-investigated field of HDS education.","abstract_html":"Health Data Science (HDS) is a new and interdisciplinary field that leverages computational methods, such as data science, statistics, and artificial intelligence, to analyse health, medical, or biological data. There is limited knowledge about how to teach and learn HDS topics effectively. Despite the urgent need to train health data scientists, educational institutions worldwide struggle to teach these interdisciplinary concepts effectively. This challenge is particularly evident in postgraduate studies, where learners come from diverse backgrounds. Additionally, there is a lack of clear guidelines on effective learning strategies in HDS to support students in addressing their educational challenges. To address this gap, this thesis aims to offer insights into health data science education through three key objectives. Firstly, it aims to uncover engagement patterns with HDS resources, including how learners interact with various types of educational resources and topics (Objective 1). Secondly, it aims to identify students’ learning tactics and strategies in health data science and explore how they impact student performance (Objective 2). Lastly, the thesis aims to develop an Artificial Intelligence (AI) assistant for the early prediction of struggling students in HDS and to provide them with personalised suggestions (Objective 3). A range of methods were employed, including qualitative content analysis, quantitative analysis of questionnaire data, and AI techniques on log data from online learning platforms. The qualitative content analysis by interviewing early career researchers from two UK institutes was conducted to understand and elaborate on the challenges they face in HDS education. Subsequent analysis then explored the self-reported learning strategies used by these researchers to overcome their educational challenges while shedding light on effective teaching methods and course design in HDS. A survey questionnaire was used to identify learning strategies and course design preferences in HDS among undergraduate and postgraduate learners across three HDS courses with varied modalities, designs, and academic levels. AI methods were used to uncover learners’ engagement with different resources as well as learning tactics and strategies based on clickstream data, reflecting the actual behaviours of learners. The thesis also analysed the correlation between the use of these learning tactics and strategies and students’ performance. Finally, a novel AI assistant was developed based on HDS students’ behaviours and learning process. The AI method used explainable AI and a counterfactual approach to provide suggestions that can help improve students’ performance. The results related to the first objective of this thesis indicate that students in HDS preferred a variety of resources and believed visually impressive materials enhance their understanding of HDS topics. Among these topics, machine learning emerged as the most interesting, while analysing genomic data was identified as the most difficult topic. Regarding the second objective, a range of learning tactics and strategies were identified, among which, connecting new information to prior knowledge and peer learning showed the highest correlation with student performance. The results also showed that students who engaged deeply with the material and applied different learning tactics throughout the course achieved higher grades than others. For the third objective, the novel AI assistant showed strong accuracy in predicting students’ performance, with an Area Under Curve (AUC) between 83% and 91%. The feedback generated by the AI, as evaluated by the course organiser, confirmed the validity of the suggestions. Additionally, the evaluation of the AI method on an external dataset outperformed traditional machine learning algorithms. The results of this thesis can potentially improve HDS education in the following ways. It can help develop effective courses through the appropriate design of learning materials (Objective 1). Also, evidence from Objective 2 can lead to guidelines that can help learners better manage their educational challenges. The AI model developed (Objective 3) can be used by teachers to identify students needing help and provide them with personalised suggestions according to their learning process. The evidence-based guidelines derived from this thesis have the potential to empower both learners and educators in HDS. We consider these to be significant contributions to the critical yet under-investigated field of HDS education.","abstract_has_math":false,"creators":["Rohani, Narges"],"institution":"The University of Edinburgh","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Gal, Yakov Kobi","Manataki, Areti","Gallagher, Michael"],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-03-04","date_published":"2025-03-04","updated_at":"2026-11-19T03:35:01Z","subjects":["learning analytics","health data science","data science","artificial intelligence","medical education","education","learning strategies"],"languages":["en"],"rights":[],"rights_urls":["https://creativecommons.org/licenses/by-nc-nd/4.0/"],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["http://dx.doi.org/10.7488/era/5708"],"render_values":[{"text":"http://dx.doi.org/10.7488/era/5708","href":"http://dx.doi.org/10.7488/era/5708","code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/1842/43167","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/1842/43167","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":"restricted","embargo_until":"2028-03-04","access_note":"Restricted access"},"source_record":{"url":"https://era.ed.ac.uk/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Aera.ed.ac.uk%3A1842%2F43167","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Gal, Yakov Kobi","Manataki, Areti","Gallagher, Michael"]},{"key":"dc:contributor.sponsor","label":"Sponsor","values":["Medical Research Council (MRC)"]},{"key":"dc:creator","label":"Author","values":["Rohani, Narges"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2025-03-04T10:08:16Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2025-03-04T10:08:16Z"]},{"key":"dc:date.issued","label":"Date","values":["2025-03-04"]},{"key":"dc:publisher","label":"Institution","values":["The University of Edinburgh"]},{"key":"dc:relation.hasversion","label":"Dc Relation Hasversion","values":["Rohani N, Sowa S, Manataki A Identifying Learning Preferences and Strategies in Health Data Science Courses: Systematic Review JMIR Medical Education 2024;10:e50667. https://doi.org/210.2196/50667","Rohani, N., Gallagher, M., Gal, K., & Manataki, A. (2024). Education Challenges Among Early Career Researchers in Medical Informatics. Studies in Health Technology and Informatics, 316, 1510–1514. https://doi.org/10.3233/SHTI240701","BMC Medical Education Journal: Rohani, N., Gallagher, M., Gal, K., & Manataki, A. (2024). Insights into health data science education: A qualitative content analysis. bioRxiv. https://doi.org/10.1101/2024.09.23.614482","Rohani, N., Gallagher, M., Gal, K., Banas, K., & Manataki, A. (2024). Self-Reported Learning Strategies and Preferences in Health Informatics. Studies in Health Technology and Informatics, 316, 1540–1544. https://doi.org/10.3233/SHTI240710","Rohani, N., Gal, K., Gallagher, M. et al. Providing insights into health data science education through artificial intelligence. BMC Medical Education 24, 564 (2024). https://doi.org/10.1186/s12909-024-05555-3","Rohani, N., Gal, K., Gallagher, M., & Manataki, A. (2023). Early prediction of student performance in a health data science MOOC. In M. Feng, T. Käser, & P. Talukdar (Eds.), Proceedings of the 16th International Conference on Educational Data Mining (EDM, 2023) (pp. 325–333). International Educational Data Mining Society. https://doi.org/10.5281/zenodo.8115721","Rohani, N., Gal, K., Gallagher, M., Manataki, A. (2023). Discovering Students’ Learning Strategies in a Visual Programming MOOC Through Process Mining Techniques. In: Montali, M., Senderovich, A., Weidlich, M. (eds) Process Mining Workshops. ICPM2022. Lecture Notes in Business Information Processing, vol 468. Springer, Cham. https://doi.org/10.1007/978-3-031-27815-0_39","Rohani, N., Rohani, B., & Manataki, A. (2024). ClickTree: A Tree-based Method for Predicting Math Students’ Performance Based on Clickstream Data. Journal of Educational Data Mining, 16(2), 32–57. https://doi.org/10.5281/zenodo.13627655."]},{"key":"dc:type","label":"Dc Type","values":["Thesis or Dissertation"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["Doctoral"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["PhD Doctor of Philosophy"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["learning analytics","health data science","data science","artificial intelligence","medical education","education","learning strategies"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]},{"key":"dc:rights.embargodate","label":"Dc Rights Embargodate","values":["2028-03-04"]},{"key":"dc:rights.license","label":"Dc Rights License","values":["CC BY-NC-ND 4.0 Attribution-NonCommercial-NoDerivatives 4.0 International Deed"]},{"key":"dc:rights.uri","label":"Rights URI","values":["https://creativecommons.org/licenses/by-nc-nd/4.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/1842/43167","http://dx.doi.org/10.7488/era/5708"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Health Data Science (HDS) is a new and interdisciplinary field that leverages computational methods, such as data science, statistics, and artificial intelligence, to analyse health, medical, or biological data. There is limited knowledge about how to teach and learn HDS topics effectively. Despite the urgent need to train health data scientists, educational institutions worldwide struggle to teach these interdisciplinary concepts effectively. This challenge is particularly evident in postgraduate studies, where learners come from diverse backgrounds. Additionally, there is a lack of clear guidelines on effective learning strategies in HDS to support students in addressing their educational challenges. To address this gap, this thesis aims to offer insights into health data science education through three key objectives. Firstly, it aims to uncover engagement patterns with HDS resources, including how learners interact with various types of educational resources and topics (Objective 1). Secondly, it aims to identify students’ learning tactics and strategies in health data science and explore how they impact student performance (Objective 2). Lastly, the thesis aims to develop an Artificial Intelligence (AI) assistant for the early prediction of struggling students in HDS and to provide them with personalised suggestions (Objective 3). A range of methods were employed, including qualitative content analysis, quantitative analysis of questionnaire data, and AI techniques on log data from online learning platforms. The qualitative content analysis by interviewing early career researchers from two UK institutes was conducted to understand and elaborate on the challenges they face in HDS education. Subsequent analysis then explored the self-reported learning strategies used by these researchers to overcome their educational challenges while shedding light on effective teaching methods and course design in HDS. A survey questionnaire was used to identify learning strategies and course design preferences in HDS among undergraduate and postgraduate learners across three HDS courses with varied modalities, designs, and academic levels. AI methods were used to uncover learners’ engagement with different resources as well as learning tactics and strategies based on clickstream data, reflecting the actual behaviours of learners. The thesis also analysed the correlation between the use of these learning tactics and strategies and students’ performance. Finally, a novel AI assistant was developed based on HDS students’ behaviours and learning process. The AI method used explainable AI and a counterfactual approach to provide suggestions that can help improve students’ performance. The results related to the first objective of this thesis indicate that students in HDS preferred a variety of resources and believed visually impressive materials enhance their understanding of HDS topics. Among these topics, machine learning emerged as the most interesting, while analysing genomic data was identified as the most difficult topic. Regarding the second objective, a range of learning tactics and strategies were identified, among which, connecting new information to prior knowledge and peer learning showed the highest correlation with student performance. The results also showed that students who engaged deeply with the material and applied different learning tactics throughout the course achieved higher grades than others. For the third objective, the novel AI assistant showed strong accuracy in predicting students’ performance, with an Area Under Curve (AUC) between 83% and 91%. The feedback generated by the AI, as evaluated by the course organiser, confirmed the validity of the suggestions. Additionally, the evaluation of the AI method on an external dataset outperformed traditional machine learning algorithms. The results of this thesis can potentially improve HDS education in the following ways. It can help develop effective courses through the appropriate design of learning materials (Objective 1). Also, evidence from Objective 2 can lead to guidelines that can help learners better manage their educational challenges. The AI model developed (Objective 3) can be used by teachers to identify students needing help and provide them with personalised suggestions according to their learning process. The evidence-based guidelines derived from this thesis have the potential to empower both learners and educators in HDS. We consider these to be significant contributions to the critical yet under-investigated field of HDS education."]},{"key":"dcterms:accessrights","label":"Dcterms Accessrights","values":["RESTRICTED ACCESS"]},{"key":"dc:title","label":"Title","values":["Supporting health data science education with learning analytics and artificial intelligence"]},{"key":"others:access_status","label":"Others Access Status","values":["embargo"]}]}],"canonical_facts":{"dc:contributor.advisor":["Gal, Yakov Kobi","Manataki, Areti","Gallagher, Michael"],"dc:contributor.sponsor":["Medical Research Council (MRC)"],"dc:creator":["Rohani, Narges"],"dc:date.accessioned":["2025-03-04T10:08:16Z"],"dc:date.available":["2025-03-04T10:08:16Z"],"dc:date.issued":["2025-03-04"],"dc:description.abstract":["Health Data Science (HDS) is a new and interdisciplinary field that leverages computational methods, such as data science, statistics, and artificial intelligence, to analyse health, medical, or biological data. There is limited knowledge about how to teach and learn HDS topics effectively. Despite the urgent need to train health data scientists, educational institutions worldwide struggle to teach these interdisciplinary concepts effectively. This challenge is particularly evident in postgraduate studies, where learners come from diverse backgrounds. Additionally, there is a lack of clear guidelines on effective learning strategies in HDS to support students in addressing their educational challenges. To address this gap, this thesis aims to offer insights into health data science education through three key objectives. Firstly, it aims to uncover engagement patterns with HDS resources, including how learners interact with various types of educational resources and topics (Objective 1). Secondly, it aims to identify students’ learning tactics and strategies in health data science and explore how they impact student performance (Objective 2). Lastly, the thesis aims to develop an Artificial Intelligence (AI) assistant for the early prediction of struggling students in HDS and to provide them with personalised suggestions (Objective 3). A range of methods were employed, including qualitative content analysis, quantitative analysis of questionnaire data, and AI techniques on log data from online learning platforms. The qualitative content analysis by interviewing early career researchers from two UK institutes was conducted to understand and elaborate on the challenges they face in HDS education. Subsequent analysis then explored the self-reported learning strategies used by these researchers to overcome their educational challenges while shedding light on effective teaching methods and course design in HDS. A survey questionnaire was used to identify learning strategies and course design preferences in HDS among undergraduate and postgraduate learners across three HDS courses with varied modalities, designs, and academic levels. AI methods were used to uncover learners’ engagement with different resources as well as learning tactics and strategies based on clickstream data, reflecting the actual behaviours of learners. The thesis also analysed the correlation between the use of these learning tactics and strategies and students’ performance. Finally, a novel AI assistant was developed based on HDS students’ behaviours and learning process. The AI method used explainable AI and a counterfactual approach to provide suggestions that can help improve students’ performance. The results related to the first objective of this thesis indicate that students in HDS preferred a variety of resources and believed visually impressive materials enhance their understanding of HDS topics. Among these topics, machine learning emerged as the most interesting, while analysing genomic data was identified as the most difficult topic. Regarding the second objective, a range of learning tactics and strategies were identified, among which, connecting new information to prior knowledge and peer learning showed the highest correlation with student performance. The results also showed that students who engaged deeply with the material and applied different learning tactics throughout the course achieved higher grades than others. For the third objective, the novel AI assistant showed strong accuracy in predicting students’ performance, with an Area Under Curve (AUC) between 83% and 91%. The feedback generated by the AI, as evaluated by the course organiser, confirmed the validity of the suggestions. Additionally, the evaluation of the AI method on an external dataset outperformed traditional machine learning algorithms. The results of this thesis can potentially improve HDS education in the following ways. It can help develop effective courses through the appropriate design of learning materials (Objective 1). Also, evidence from Objective 2 can lead to guidelines that can help learners better manage their educational challenges. The AI model developed (Objective 3) can be used by teachers to identify students needing help and provide them with personalised suggestions according to their learning process. The evidence-based guidelines derived from this thesis have the potential to empower both learners and educators in HDS. We consider these to be significant contributions to the critical yet under-investigated field of HDS education."],"dc:identifier.uri":["https://hdl.handle.net/1842/43167","http://dx.doi.org/10.7488/era/5708"],"dc:language.iso":["en"],"dc:publisher":["The University of Edinburgh"],"dc:relation.hasversion":["Rohani N, Sowa S, Manataki A Identifying Learning Preferences and Strategies in Health Data Science Courses: Systematic Review JMIR Medical Education 2024;10:e50667. https://doi.org/210.2196/50667","Rohani, N., Gallagher, M., Gal, K., & Manataki, A. (2024). Education Challenges Among Early Career Researchers in Medical Informatics. Studies in Health Technology and Informatics, 316, 1510–1514. https://doi.org/10.3233/SHTI240701","BMC Medical Education Journal: Rohani, N., Gallagher, M., Gal, K., & Manataki, A. (2024). Insights into health data science education: A qualitative content analysis. bioRxiv. https://doi.org/10.1101/2024.09.23.614482","Rohani, N., Gallagher, M., Gal, K., Banas, K., & Manataki, A. (2024). Self-Reported Learning Strategies and Preferences in Health Informatics. Studies in Health Technology and Informatics, 316, 1540–1544. https://doi.org/10.3233/SHTI240710","Rohani, N., Gal, K., Gallagher, M. et al. Providing insights into health data science education through artificial intelligence. BMC Medical Education 24, 564 (2024). https://doi.org/10.1186/s12909-024-05555-3","Rohani, N., Gal, K., Gallagher, M., & Manataki, A. (2023). Early prediction of student performance in a health data science MOOC. In M. Feng, T. Käser, & P. Talukdar (Eds.), Proceedings of the 16th International Conference on Educational Data Mining (EDM, 2023) (pp. 325–333). International Educational Data Mining Society. https://doi.org/10.5281/zenodo.8115721","Rohani, N., Gal, K., Gallagher, M., Manataki, A. (2023). Discovering Students’ Learning Strategies in a Visual Programming MOOC Through Process Mining Techniques. In: Montali, M., Senderovich, A., Weidlich, M. (eds) Process Mining Workshops. ICPM2022. Lecture Notes in Business Information Processing, vol 468. Springer, Cham. https://doi.org/10.1007/978-3-031-27815-0_39","Rohani, N., Rohani, B., & Manataki, A. (2024). ClickTree: A Tree-based Method for Predicting Math Students’ Performance Based on Clickstream Data. Journal of Educational Data Mining, 16(2), 32–57. https://doi.org/10.5281/zenodo.13627655."],"dc:rights.embargodate":["2028-03-04"],"dc:rights.license":["CC BY-NC-ND 4.0 Attribution-NonCommercial-NoDerivatives 4.0 International Deed"],"dc:rights.uri":["https://creativecommons.org/licenses/by-nc-nd/4.0/"],"dc:subject":["learning analytics","health data science","data science","artificial intelligence","medical education","education","learning strategies"],"dcterms:accessrights":["RESTRICTED ACCESS"],"dc:title":["Supporting health data science education with learning analytics and artificial intelligence"],"dc:type":["Thesis or Dissertation"],"dc:type.qualificationlevel":["Doctoral"],"dc:type.qualificationname":["PhD Doctor of Philosophy"],"others:access_status":["embargo"]},"updated_at":"2026-11-19T03:35:01Z"}