Liesbet M. Peeters
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Liesbet M. Peeters has a total of 405 co-authors. Here are the top co-authors by number of shared publications:
Bart Van Wijmeersch
University MS Centre (UMSC)
14 shared publications- Nadine Buys
Center for Animal Breeding and Genetics, Department of Biosystems, KU Leuven, Leuven, Belgium
12 shared publications - Ashkan Pirmani
ESAT, STADIUS, KU Leuven, Leuven, Belgium
9 shared publications - Yves Moreau
Department of Electrical Engineering, KU Leuven, Leuven
8 shared publications - Edward De Brouwer
ESAT-STADIUS KU Leuven
7 shared publications Edward De Brouwer
ESAT-STADIUS, KU Leuven
6 shared publications
Latest Publications
- Open Access
- 5 June 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Piloting the automatic transformation of real-world data to the OMOP common data model: the MS core dataset to OMOP SwitchBox (Preprint)
BACKGROUND Real-world data (RWD) gathered from multiple sclerosis (MS) registries and cohorts is essential for deepening our understanding of this complex autoimmune disease. However, the transition from RWD to real-world evidence (RWE) faces challenges due to substantial data heterogeneity arising from varying dataset definitions and varying local compliance with global standards. This inconsistency constrains the comparability and applicability of data across various studies and settings. OBJECTIVE This paper presents the pilot of a tool, named "SwitchBox," to facilitate and streamline the conversion of MS RWD into the Observational Medical Outcomes Partnership Common Data Model (OMOP CDM), enhancing data harmonisation and enabling easier collaboration across healthcare datasets. METHODS The SwitchBox is a framework for ETL (Extract-Transform-Load) modules that transform data in a specific format into the OMOP CDM. The MS Data Alliance (MSDA) Core Dataset (CDS) acts as the pilot ETL module. In a collaboration between academia and industry, the development focused on two major parts: (1) the framework and (2) the MSDA CDS ETL development. The mapping process followed the basic recommendations for ETL development provided by EHDEN/OHDSI, including the development of a synthetic dataset based on MSDA CDS data dictionary without real patient information to ensure data privacy, the creation of a scan report using the OHDSI tool "White Rabbit" and the utilisation of "Rabbit in a Hat" for developing a structural mapping document, aligning source variables with OMOP CDM tables and concept IDs via "Athena." RESULTS The developed SwitchBox successfully transforms data in the MSDA CDS format to the OMOP CDM. The mapping process turned out to be the most challenging aspect of the SwitchBox project. For the 44 MSDA CDS variables, we targeted eight OMOP CDM tables, with the Observation table being the most utilised. We identified specific challenges in the structural mapping process, which were not unique to MS data but could inform other registry-based sources. The SwitchBox pilot is distributed as a Docker container and is provided via a GitHub container registry. CONCLUSIONS The development of the SwitchBox represents a significant advancement in the standardisation of MS RWD collection and analysis. By simplifying the conversion to the OMOP CDM, this tool mitigates barriers to data harmonisation and facilitates broader collaborations across healthcare research. The findings highlight the need to integrate user-friendly solutions to transform RWD into actionable insights, underscoring the potential for greater data utility in clinical, research, or policy-making contexts. Further research should focus on expanding the tool’s application to a broader array of data sources and disease areas.
- 5 June 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessPiloting the automatic transformation of real-world data to the OMOP common data model: the MS core dataset to OMOP SwitchBox (Preprint)
BACKGROUND Real-world data (RWD) gathered from multiple sclerosis (MS) registries and cohorts is essential for deepening our understanding of this complex autoimmune disease. However, the transition from RWD to real-world evidence (RWE) faces challenges due to substantial data heterogeneity arising from varying dataset definitions and varying local compliance with global standards. This inconsistency constrains the comparability and applicability of data across various studies and settings. OBJECTIVE This paper presents the pilot of a tool, named "SwitchBox," to facilitate and streamline the conversion of MS RWD into the Observational Medical Outcomes Partnership Common Data Model (OMOP CDM), enhancing data harmonisation and enabling easier collaboration across healthcare datasets. METHODS The SwitchBox is a framework for ETL (Extract-Transform-Load) modules that transform data in a specific format into the OMOP CDM. The MS Data Alliance (MSDA) Core Dataset (CDS) acts as the pilot ETL module. In a collaboration between academia and industry, the development focused on two major parts: (1) the framework and (2) the MSDA CDS ETL development. The mapping process followed the basic recommendations for ETL development provided by EHDEN/OHDSI, including the development of a synthetic dataset based on MSDA CDS data dictionary without real patient information to ensure data privacy, the creation of a scan report using the OHDSI tool "White Rabbit" and the utilisation of "Rabbit in a Hat" for developing a structural mapping document, aligning source variables with OMOP CDM tables and concept IDs via "Athena." RESULTS The developed SwitchBox successfully transforms data in the MSDA CDS format to the OMOP CDM. The mapping process turned out to be the most challenging aspect of the SwitchBox project. For the 44 MSDA CDS variables, we targeted eight OMOP CDM tables, with the Observation table being the most utilised. We identified specific challenges in the structural mapping process, which were not unique to MS data but could inform other registry-based sources. The SwitchBox pilot is distributed as a Docker container and is provided via a GitHub container registry. CONCLUSIONS The development of the SwitchBox represents a significant advancement in the standardisation of MS RWD collection and analysis. By simplifying the conversion to the OMOP CDM, this tool mitigates barriers to data harmonisation and facilitates broader collaborations across healthcare research. The findings highlight the need to integrate user-friendly solutions to transform RWD into actionable insights, underscoring the potential for greater data utility in clinical, research, or policy-making contexts. Further research should focus on expanding the tool’s application to a broader array of data sources and disease areas.
- Open Access
- 11 March 2026
- Published by Frontiers Media SA in Frontiers in Immunology
- Vol. 17, 1625837
Combining magnetic resonance imaging and evoked potentials enhances machine learning prediction of multiple sclerosis disability worsening
Introduction: Predicting long-term disability progression in multiple sclerosis (MS) remains a significant challenge. Existing prognostic models often rely on single-modality data or conventional measures, such as lesion count on magnetic resonance imaging (MRI) or latency values from evoked potentials (EPs), overlooking subclinical disease progression. This study aimed to develop a multimodal machine learning (ML) pipeline integrating clinical, high-dimensional MRI, and motor EP time-series (EPTS) features to predict disability worsening in MS. Methods: A retrospective cohort of 127 people with MS (PwMS; 424 episodes) from a tertiary MS center in Belgium was used, including clinical data, T2-weighted fluid-attenuated inversion recovery MRI, and motor EPs. Disability worsening was defined as a change in the expanded disability status scale (EDSS) over two years, stratified by baseline EDSS. MRI features included 42 anatomical and lesion volumes and 100 radiomic descriptors from lesions and the normal-appearing white matter (NAWM). EPTS features included latency, peak-to-peak amplitude (PPA), and high-dimensional descriptors selected using highly comparative time-series analysis (HCTSA) and Boruta. ML models (Light Gradient Boosting Machine (LGBM), random forest, logistic regression) were trained using 20×repeated stratified 3-fold cross-validation. Performance was evaluated using the area under the receiver operating characteristic curve (AUROC), average precision (AP), and Brier score. SHapley Additive exPlanations (SHAP) were used for interpretability. Results: Across 96 model configurations, models combining MRI and EPTS features, with or without clinical data, consistently outperformed single-modality models across AUROC, AP, and Brier score, regardless of algorithm or feature representation. The best-performing model (Brier score = 0.062) was an LGBM using combined MRI and EPTS data. MRI radiomics dominated feature importance, especially shape- and texture-based features from NAWM and lesion regions. EPTS features, particularly waveform dynamics (e.g., Sliding Window) and PPA, provided complementary value and improved sensitivity. EPTS-only models showed the highest AUROC, but combined models achieved the best overall balance across all performance metrics. Conclusion: This is the first study to integrate clinical, MRI radiomics, and motor EPTS features in an ML pipeline for MS prognosis. Combining structural and functional subclinical markers improves the prediction of disability worsening and supports multimodal monitoring for personalized care.
- 11 March 2026
- Published by Frontiers Media SA in Frontiers in Immunology
- Vol. 17, 1625837
Open AccessCombining magnetic resonance imaging and evoked potentials enhances machine learning prediction of multiple sclerosis disability worsening
Introduction: Predicting long-term disability progression in multiple sclerosis (MS) remains a significant challenge. Existing prognostic models often rely on single-modality data or conventional measures, such as lesion count on magnetic resonance imaging (MRI) or latency values from evoked potentials (EPs), overlooking subclinical disease progression. This study aimed to develop a multimodal machine learning (ML) pipeline integrating clinical, high-dimensional MRI, and motor EP time-series (EPTS) features to predict disability worsening in MS. Methods: A retrospective cohort of 127 people with MS (PwMS; 424 episodes) from a tertiary MS center in Belgium was used, including clinical data, T2-weighted fluid-attenuated inversion recovery MRI, and motor EPs. Disability worsening was defined as a change in the expanded disability status scale (EDSS) over two years, stratified by baseline EDSS. MRI features included 42 anatomical and lesion volumes and 100 radiomic descriptors from lesions and the normal-appearing white matter (NAWM). EPTS features included latency, peak-to-peak amplitude (PPA), and high-dimensional descriptors selected using highly comparative time-series analysis (HCTSA) and Boruta. ML models (Light Gradient Boosting Machine (LGBM), random forest, logistic regression) were trained using 20×repeated stratified 3-fold cross-validation. Performance was evaluated using the area under the receiver operating characteristic curve (AUROC), average precision (AP), and Brier score. SHapley Additive exPlanations (SHAP) were used for interpretability. Results: Across 96 model configurations, models combining MRI and EPTS features, with or without clinical data, consistently outperformed single-modality models across AUROC, AP, and Brier score, regardless of algorithm or feature representation. The best-performing model (Brier score = 0.062) was an LGBM using combined MRI and EPTS data. MRI radiomics dominated feature importance, especially shape- and texture-based features from NAWM and lesion regions. EPTS features, particularly waveform dynamics (e.g., Sliding Window) and PPA, provided complementary value and improved sensitivity. EPTS-only models showed the highest AUROC, but combined models achieved the best overall balance across all performance metrics. Conclusion: This is the first study to integrate clinical, MRI radiomics, and motor EPTS features in an ML pipeline for MS prognosis. Combining structural and functional subclinical markers improves the prediction of disability worsening and supports multimodal monitoring for personalized care.
- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 324-331
Applying Federated Learning to Block-Term Tensor Regression for Decentralised Data Analysis of Biomedical Data
Block-Term Tensor Regression (BTTR) is a powerful method for modeling complex, high-dimensional data through multilinear relationships, making it well-suited for healthcare and neuroscience. However, its reliance on centralized datasets raises privacy concerns and limits collaboration. To address this, we propose Federated Block-Term Tensor Regression (FBTTR), an extension of BTTR for federated learning that enables decentralized model building while preserving privacy and regulatory compliance. We evaluate FBTTR in two case studies: finger movement decoding from Electrocorticography (ECoG) signals and disease risk prediction. On the BCI Competition IV dataset, FBTTR outperforms non-multilinear models and achieves higher accuracy than centralized BTTR (e.g., subject 3 thumb decoding: $0.76 \pm 0.05$ vs. $0.71 \pm 0.05$). On real-world clinical data, FBTTR surpasses both standard federated approaches and centralized BTTR (e.g., Fed-Heart-Disease Dataset AUC-ROC: $0.872 \pm 0.02$ vs. $0.812 \pm 0.003$; accuracy: $0.772 \pm 0.02$ vs. $0.753 \pm 0.007$). These results show that FBTTR is scalable, computationally efficient, and achieves predictive performance comparable to centralized models. Released as open-source software, it provides a practical and transparent framework for advancing federated analytics in healthcare and brain-computer interface applications.- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 324-331
Applying Federated Learning to Block-Term Tensor Regression for Decentralised Data Analysis of Biomedical Data
Block-Term Tensor Regression (BTTR) is a powerful method for modeling complex, high-dimensional data through multilinear relationships, making it well-suited for healthcare and neuroscience. However, its reliance on centralized datasets raises privacy concerns and limits collaboration. To address this, we propose Federated Block-Term Tensor Regression (FBTTR), an extension of BTTR for federated learning that enables decentralized model building while preserving privacy and regulatory compliance. We evaluate FBTTR in two case studies: finger movement decoding from Electrocorticography (ECoG) signals and disease risk prediction. On the BCI Competition IV dataset, FBTTR outperforms non-multilinear models and achieves higher accuracy than centralized BTTR (e.g., subject 3 thumb decoding: $0.76 \pm 0.05$ vs. $0.71 \pm 0.05$). On real-world clinical data, FBTTR surpasses both standard federated approaches and centralized BTTR (e.g., Fed-Heart-Disease Dataset AUC-ROC: $0.872 \pm 0.02$ vs. $0.812 \pm 0.003$; accuracy: $0.772 \pm 0.02$ vs. $0.753 \pm 0.007$). These results show that FBTTR is scalable, computationally efficient, and achieves predictive performance comparable to centralized models. Released as open-source software, it provides a practical and transparent framework for advancing federated analytics in healthcare and brain-computer interface applications.- Open Access
- 29 September 2025
- Published by Springer Nature in Journal of NeuroEngineering and Rehabilitation
- Vol. 22 (1) , 1-11
Integrating big data and artificial intelligence to predict progression in multiple sclerosis: challenges and the path forward
Multiple sclerosis (MS) remains a complex and costly neurological condition characterised by progressive disability, making early detection and accurate prognosis of disease progression imperative. While artificial intelligence (AI) combined with big data promises transformative advances in personalised MS care, integration of multimodal, real-world datasets, including clinical records, magnetic resonance imaging (MRI), and digital biomarkers, remains limited. This perspective paper identifies a critical gap between technical innovation and clinical implementation, driven by methodological constraints, evolving regulatory frameworks, and ethical concerns related to bias, privacy, and equity. We explore this gap through three interconnected lenses: the underuse of integrated real-world data, the barriers posed by regulation and ethics, and emerging solutions. Promising strategies such as federated learning, regulatory initiatives like DARWIN-EU and the European Health Data Space, and patient-led frameworks including PROMS and CLAIMS, offer structured pathways forward. Additionally, we highlight the growing relevance of foundation models for interpreting complex MS data and supporting clinical decision-making. We advocate for harmonised data infrastructures, patient-centred design, explainable AI, and real-world validation as core pillars for future implementation. By aligning technical, regulatory, and ethical domains, stakeholders can unlock the full potential of AI to enhance prognosis, personalise care, and improve outcomes for people with MS.
- 29 September 2025
- Published by Springer Nature in Journal of NeuroEngineering and Rehabilitation
- Vol. 22 (1) , 1-11
Open AccessIntegrating big data and artificial intelligence to predict progression in multiple sclerosis: challenges and the path forward
Multiple sclerosis (MS) remains a complex and costly neurological condition characterised by progressive disability, making early detection and accurate prognosis of disease progression imperative. While artificial intelligence (AI) combined with big data promises transformative advances in personalised MS care, integration of multimodal, real-world datasets, including clinical records, magnetic resonance imaging (MRI), and digital biomarkers, remains limited. This perspective paper identifies a critical gap between technical innovation and clinical implementation, driven by methodological constraints, evolving regulatory frameworks, and ethical concerns related to bias, privacy, and equity. We explore this gap through three interconnected lenses: the underuse of integrated real-world data, the barriers posed by regulation and ethics, and emerging solutions. Promising strategies such as federated learning, regulatory initiatives like DARWIN-EU and the European Health Data Space, and patient-led frameworks including PROMS and CLAIMS, offer structured pathways forward. Additionally, we highlight the growing relevance of foundation models for interpreting complex MS data and supporting clinical decision-making. We advocate for harmonised data infrastructures, patient-centred design, explainable AI, and real-world validation as core pillars for future implementation. By aligning technical, regulatory, and ethical domains, stakeholders can unlock the full potential of AI to enhance prognosis, personalise care, and improve outcomes for people with MS.
- Open Access
- 20 August 2025
- Published by Frontiers Media SA in Frontiers in Immunology
- Vol. 16, 1679482
Editorial: Use of big data and artificial intelligence in multiple sclerosis
As health data volume and the sophistication of artificial intelligence (AI) tools grow, their potential to transform the management of complex neuroimmunological conditions like multiple sclerosis (MS) has become increasingly evident (Schwalbe et al. 2020). MS, a chronic immune-mediated inflammatory disorder of the central nervous system, presents a unique challenge in the health sector due to its multifactorial nature and variable progression patterns. Each patient's journey is marked by distinct symptom trajectories and responses to treatment, demanding personalised approaches in diagnosis, prognosis, and therapeutic interventions. (Kuhlmann T et al. 2023, Jakimowski et al, 2024). We believe this special topic has opened new perspectives, and gives us some indications of where the field of Big Data and AI in MS is heading. First of all, it testifies that the domain is expanding rapidly. At the same time, however, researchers will have to solve some open issues, such as the need to develop trustworthy, reliable AI models, consistently capture multidimensional longitudinal data, incorporate the patient perspectives and the alignment with evolving regulatory frameworks such as the EHDS. We hope you find this collection as inspiring and impactful to read as it was for us to prepare.
- 20 August 2025
- Published by Frontiers Media SA in Frontiers in Immunology
- Vol. 16, 1679482
Open AccessEditorial: Use of big data and artificial intelligence in multiple sclerosis
As health data volume and the sophistication of artificial intelligence (AI) tools grow, their potential to transform the management of complex neuroimmunological conditions like multiple sclerosis (MS) has become increasingly evident (Schwalbe et al. 2020). MS, a chronic immune-mediated inflammatory disorder of the central nervous system, presents a unique challenge in the health sector due to its multifactorial nature and variable progression patterns. Each patient's journey is marked by distinct symptom trajectories and responses to treatment, demanding personalised approaches in diagnosis, prognosis, and therapeutic interventions. (Kuhlmann T et al. 2023, Jakimowski et al, 2024). We believe this special topic has opened new perspectives, and gives us some indications of where the field of Big Data and AI in MS is heading. First of all, it testifies that the domain is expanding rapidly. At the same time, however, researchers will have to solve some open issues, such as the need to develop trustworthy, reliable AI models, consistently capture multidimensional longitudinal data, incorporate the patient perspectives and the alignment with evolving regulatory frameworks such as the EHDS. We hope you find this collection as inspiring and impactful to read as it was for us to prepare.
- 2 July 2025
- Published by Springer Nature in The VLDB Journal
- Vol. 34 (4) , 1-27
Measuring approximate functional dependencies: a comparative study
- 2 July 2025
- Published by Springer Nature in The VLDB Journal
- Vol. 34 (4) , 1-27
Measuring approximate functional dependencies: a comparative study
- Open Access
- 22 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 203-206
LLM-M atcher : A Name-Based Schema Matching Tool using Large Language Models
- 22 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 203-206
Open AccessLLM-M atcher : A Name-Based Schema Matching Tool using Large Language Models
- Open Access
An Algorithmic Pipeline for GDPR-Compliant Healthcare Data Anonymisation: Moving Toward Standardisation
High-quality real-world data (RWD) is essential for healthcare but must be transformed to comply with the General Data Protection Regulation (GDPR). GDPRs broad definitions of quasi-identifiers (QIDs) and sensitive attributes (SAs) complicate implementation. We aim to standardise RWD anonymisation for GDPR compliance while preserving data utility by introducing an algorithmic method to identify QIDs and SAs and evaluate utility in anonymised datasets. We conducted a systematic literature review via ProQuest and PubMed to inform a three-stage anonymisation pipeline: identification, de-identification, and quasi-identifier dimension evaluation. The pipeline was implemented, validated, and tested on two mock RWD datasets (500 and 1000 rows). Privacy was assessed using k-anonymity, l-diversity, and t-closeness; utility was measured by non-uniform entropy (NUE). The review yielded two studies on QID/SA identification and five on utility metrics. Applying the pipeline, attributes were classified by re-identification risk using alpha and beta thresholds (25 percent/1 percent for 500 rows; 10 percent/1 percent for 1000 rows). Privacy metrics improved k-anonymity from 1 to 4 (500 rows) and 1 to 110 (1000 rows). NUE scores were 69.26 percent and 69.05 percent, respectively, indicating consistent utility despite varying privacy gains. We present a GDPR-compliant anonymisation pipeline for healthcare RWD that provides a reproducible approach to QID/SA identification and utility evaluation; publicly available code promotes standardisation, data privacy, and open science.
Open AccessAn Algorithmic Pipeline for GDPR-Compliant Healthcare Data Anonymisation: Moving Toward Standardisation
High-quality real-world data (RWD) is essential for healthcare but must be transformed to comply with the General Data Protection Regulation (GDPR). GDPRs broad definitions of quasi-identifiers (QIDs) and sensitive attributes (SAs) complicate implementation. We aim to standardise RWD anonymisation for GDPR compliance while preserving data utility by introducing an algorithmic method to identify QIDs and SAs and evaluate utility in anonymised datasets. We conducted a systematic literature review via ProQuest and PubMed to inform a three-stage anonymisation pipeline: identification, de-identification, and quasi-identifier dimension evaluation. The pipeline was implemented, validated, and tested on two mock RWD datasets (500 and 1000 rows). Privacy was assessed using k-anonymity, l-diversity, and t-closeness; utility was measured by non-uniform entropy (NUE). The review yielded two studies on QID/SA identification and five on utility metrics. Applying the pipeline, attributes were classified by re-identification risk using alpha and beta thresholds (25 percent/1 percent for 500 rows; 10 percent/1 percent for 1000 rows). Privacy metrics improved k-anonymity from 1 to 4 (500 rows) and 1 to 110 (1000 rows). NUE scores were 69.26 percent and 69.05 percent, respectively, indicating consistent utility despite varying privacy gains. We present a GDPR-compliant anonymisation pipeline for healthcare RWD that provides a reproducible approach to QID/SA identification and utility evaluation; publicly available code promotes standardisation, data privacy, and open science.
- Open Access
Developing a Population Health Management Dashboard for Belgium: a focus group study with mock-up design
Background: The necessary data for delivering targeted population health management (PHM) interventions in Belgium is insufficient and scattered due to limited, non- interoperable data sources. Insights on data availability, user needs, and analysis capabilities should be obtained to develop a Belgian PHM dashboard. We aimed to identify visual, content, and access requirements for a PHM dashboard, and visualize user needs through a mock-up dashboard to enhance quintuple aim. Methods: Five focus groups were conducted in three Belgian locoregional health networks (collaborations of primary care zones), comprising 35 potential future dashboard users with various backgrounds. Insights were gathered using open-ended questions based on frameworks of successful international examples. These insights were incorporated into a mock-up dashboard. Subsequently, the mock-up dashboard was discussed in a second round of focus groups. Data from the focus groups were analysed using the Qualitative Analysis Guide of Leuven (QUAGOL) method. Results: Differentiated user requirements can be classified into: (1) ’General Exploration’ allowing for assessing the health and socioeconomic status and risk factors in a region using PHM indicators, (2) ‘Risk Stratification and Selection of Interventions’ enabling matching of resources and interventions to at-risk subpopulations, (3) ‘Research Community’ providing advanced tools for data exploration and collaboration with the research community. The mock-up dashboard meets these requirements according to respondents, by supporting population health managers, healthcare professionals and policymakers with easy access to insights and expert population health managers, such as researchers and data analysts, with options for advanced data analysis. Conclusions: Our study identified ‘exploration’, ‘risk stratification’, ‘intervention selection’ and ‘advanced data analysis’ as key components of an interoperable and customizable dashboard, contributing to a more efficient and equitable healthcare system. The developed PHM mock-up dashboard encapsulates these features, aiming to achieve the quintuple aim in Belgium.
Open AccessDeveloping a Population Health Management Dashboard for Belgium: a focus group study with mock-up design
Background: The necessary data for delivering targeted population health management (PHM) interventions in Belgium is insufficient and scattered due to limited, non- interoperable data sources. Insights on data availability, user needs, and analysis capabilities should be obtained to develop a Belgian PHM dashboard. We aimed to identify visual, content, and access requirements for a PHM dashboard, and visualize user needs through a mock-up dashboard to enhance quintuple aim. Methods: Five focus groups were conducted in three Belgian locoregional health networks (collaborations of primary care zones), comprising 35 potential future dashboard users with various backgrounds. Insights were gathered using open-ended questions based on frameworks of successful international examples. These insights were incorporated into a mock-up dashboard. Subsequently, the mock-up dashboard was discussed in a second round of focus groups. Data from the focus groups were analysed using the Qualitative Analysis Guide of Leuven (QUAGOL) method. Results: Differentiated user requirements can be classified into: (1) ’General Exploration’ allowing for assessing the health and socioeconomic status and risk factors in a region using PHM indicators, (2) ‘Risk Stratification and Selection of Interventions’ enabling matching of resources and interventions to at-risk subpopulations, (3) ‘Research Community’ providing advanced tools for data exploration and collaboration with the research community. The mock-up dashboard meets these requirements according to respondents, by supporting population health managers, healthcare professionals and policymakers with easy access to insights and expert population health managers, such as researchers and data analysts, with options for advanced data analysis. Conclusions: Our study identified ‘exploration’, ‘risk stratification’, ‘intervention selection’ and ‘advanced data analysis’ as key components of an interoperable and customizable dashboard, contributing to a more efficient and equitable healthcare system. The developed PHM mock-up dashboard encapsulates these features, aiming to achieve the quintuple aim in Belgium.
- Open Access
- 10 April 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Data Management Approaches for Health Data Integration: a Literature Review (Preprint)
BACKGROUND Data integration describes transformation processes that allow users to uniformly access a set of autonomous and heterogeneous datasets. Data integration pipelines consist of multiple data engineering tasks (data profiling, schema matching and mapping, entity matching, and data mapping), each of which is complex and time-consuming. Given the amount and heterogeneity of real-world data, (semi-)automated approaches to solve data engineering tasks can potentially accelerate the development of health data integration pipelines. In data management, researchers developed many such automations. It remains unclear, however, whether automated data integration approaches developed in data management can be applied to the healthcare domain. OBJECTIVE With this study, we investigate whether automated approaches for data engineering tasks from data management are applied in medical informatics. We study which data engineering tasks are automated and what kind of automation is applied. As we see a large potential for collaboration, we address researchers from both data management and medical informatics in order to promote collaboration between these two research fields. METHODS We perform a literature review of all publications that describe health data integration or solving a health data engineering task while referencing papers from data management. RESULTS From a total of 18423 papers we identified 89 that automate a health data engineering task. Most commonly, medical informaticians automate entity matching, followed by schema matching and mapping. Data profiling and data mapping are less present. The most common automation techniques are string similarity assessments and machine learning approaches. CONCLUSIONS We found positive examples of automated health data integration in our literature review. However, given the large number of manual approaches to solving data integration we found, there remains room for improvement with respect to the adoption of automated integrations. Solutions from data management have great potential in accelerating automated health data integration. With this review, we provide researchers from medical informatics with an overview of recent automation approaches developed in data management. Furthermore, we highlight the adoption of automated approaches in healthcare, clarifying the health data integration challenges to researchers from data management. This literature review serves as an entry point to data management and medical informatics researchers, facilitating a higher understanding of both domains.
- 10 April 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessData Management Approaches for Health Data Integration: a Literature Review (Preprint)
BACKGROUND Data integration describes transformation processes that allow users to uniformly access a set of autonomous and heterogeneous datasets. Data integration pipelines consist of multiple data engineering tasks (data profiling, schema matching and mapping, entity matching, and data mapping), each of which is complex and time-consuming. Given the amount and heterogeneity of real-world data, (semi-)automated approaches to solve data engineering tasks can potentially accelerate the development of health data integration pipelines. In data management, researchers developed many such automations. It remains unclear, however, whether automated data integration approaches developed in data management can be applied to the healthcare domain. OBJECTIVE With this study, we investigate whether automated approaches for data engineering tasks from data management are applied in medical informatics. We study which data engineering tasks are automated and what kind of automation is applied. As we see a large potential for collaboration, we address researchers from both data management and medical informatics in order to promote collaboration between these two research fields. METHODS We perform a literature review of all publications that describe health data integration or solving a health data engineering task while referencing papers from data management. RESULTS From a total of 18423 papers we identified 89 that automate a health data engineering task. Most commonly, medical informaticians automate entity matching, followed by schema matching and mapping. Data profiling and data mapping are less present. The most common automation techniques are string similarity assessments and machine learning approaches. CONCLUSIONS We found positive examples of automated health data integration in our literature review. However, given the large number of manual approaches to solving data integration we found, there remains room for improvement with respect to the adoption of automated integrations. Solutions from data management have great potential in accelerating automated health data integration. With this review, we provide researchers from medical informatics with an overview of recent automation approaches developed in data management. Furthermore, we highlight the adoption of automated approaches in healthcare, clarifying the health data integration challenges to researchers from data management. This literature review serves as an entry point to data management and medical informatics researchers, facilitating a higher understanding of both domains.