Periodic Reporting for period 2 - RES-Q PLUS (Comprehensive solutions of healthcare improvement based on the global Registry of Stroke Care Quality)
Berichtszeitraum: 2024-05-01 bis 2025-10-31
The primary objective of the RES-Q+ project is to improve the collection, analysis and use of data on stroke care provided by hospitals by capturing the entire patient care pathway. The project responds to current healthcare challenges, including staff shortages, ageing populations and increasing demands on healthcare systems, by supporting systematic monitoring and improvement of stroke care quality.
RES-Q+ follows the RES-Q Registry in Stroke Care Quality, a global programme for monitoring and improving stroke care quality, developed and conducted in cooperation with key stakeholders - European Stroke Organisation, World Stroke Organization and ANGELS Initiative. The project aims to enhance stroke care quality and patient outcomes through advanced data collection, semantic harmonisation and analysis of clinical data.
The consortium applies Natural Language Processing (NLP) and a clinically validated semantic model to automate the ingestion of hospital discharge reports across multiple languages. These technologies support audit and feedback processes, enable standardisation of discharge reports, assist in imputing missing data and reduce the burden of manual data entry for healthcare professionals.
RES-Q+ also develops two AI-enabled virtual assistants: one supporting patients in reporting outcomes and feedback after discharge, and one supporting clinicians in accessing and analysing stroke care quality data.
Social sciences and humanities are incorporated through user-centred design research to understand patient and clinician needs and behaviour. Ethical and legal considerations are embedded throughout the development and deployment of all project solutions.
RES-Q+ follows the RES-Q Registry in Stroke Care Quality, a global programme for monitoring and improving stroke care quality, developed and conducted in cooperation with key stakeholders - European Stroke Organisation, World Stroke Organization and ANGELS Initiative. The project aims to enhance stroke care quality and patient outcomes through advanced data collection, semantic harmonisation and analysis of clinical data.
The consortium applies Natural Language Processing (NLP) and a clinically validated semantic model to automate the ingestion of hospital discharge reports across multiple languages. These technologies support audit and feedback processes, enable standardisation of discharge reports, assist in imputing missing data and reduce the burden of manual data entry for healthcare professionals.
RES-Q+ also develops two AI-enabled virtual assistants: one supporting patients in reporting outcomes and feedback after discharge, and one supporting clinicians in accessing and analysing stroke care quality data.
Social sciences and humanities are incorporated through user-centred design research to understand patient and clinician needs and behaviour. Ethical and legal considerations are embedded throughout the development and deployment of all project solutions.
During the reporting period, the RES-Q+ consortium carried out activities across design, technical development, validation, legal and ethical compliance, and system integration.
A major achievement is the development and deployment of a tool for generating standardised stroke hospital discharge reports directly within the RES-Q platform. Based on analyses of discharge reports from several European countries and extensive clinician engagement, the Discharge Report Composer enables the automatic generation of discharge reports with standardised content and structure in multiple languages. The tool is fully operational and available to RES-Q users.
In parallel, the interim version of the Discharge Importer Tool (DIT) has been developed and integrated with RES-Q. Using the DIT, a predefined set of users can extract data from discharge reports and transfer it into RES-Q. This tool uses NLP methods to extract relevant clinical information from unstructured hospital discharge reports and allows medical experts to validate the extracted data before submission.
The consortium also designed and developed interim versions of virtual assistants for patients and clinicians using a participatory and iterative design approach. The Patient Virtual Assistant (pVA) supports the collection of patient-reported outcome measures after discharge and has entered pilot clinical evaluation. The Clinician Virtual Assistant (cVA) enables interaction with stroke care quality data and dashboards and has undergone initial usability testing. The final versions of the pVA and cVA have been designed and are currently being implemented by the technical partners and will be released for testing and evaluation in the Spring of 2026.
Predictive modeling capabilities were developed, including a fully functional Data Science Workbench integrating visualization and predictive tools, and machine learning models for predicting 90-day stroke outcomes and post-stroke conditions, all enhanced by explainable AI features to support clinical trust and transparency. The predictive tool is currently being evaluated with clinicians with synthetic data and could be made available to RES-Q users once MDR approval has been obtained.
A semantic harmonization layer has been developed, incorporating FAIRification infrastructure, a Data Catalog, and ontology mapping through knowledge graphs. The data model has been aligned with standardized clinical terminologies, ensuring interoperability, consistent data representation, and facilitating data integration and reuse across systems, e.g. other registries. Given appropriate approval, future studies could pool data from multiple registries for e.g. analysis of care quality data, predictions of outcomes.
All developments were accompanied by continuous legal and ethical assessment, including data protection impact assessments, data sharing agreements and analysis of regulatory requirements relevant to data processing, artificial intelligence and medical devices.
Figure 2: The RES-Q+ consortium at the M30 project meeting in Murcia, 2025.
A major achievement is the development and deployment of a tool for generating standardised stroke hospital discharge reports directly within the RES-Q platform. Based on analyses of discharge reports from several European countries and extensive clinician engagement, the Discharge Report Composer enables the automatic generation of discharge reports with standardised content and structure in multiple languages. The tool is fully operational and available to RES-Q users.
In parallel, the interim version of the Discharge Importer Tool (DIT) has been developed and integrated with RES-Q. Using the DIT, a predefined set of users can extract data from discharge reports and transfer it into RES-Q. This tool uses NLP methods to extract relevant clinical information from unstructured hospital discharge reports and allows medical experts to validate the extracted data before submission.
The consortium also designed and developed interim versions of virtual assistants for patients and clinicians using a participatory and iterative design approach. The Patient Virtual Assistant (pVA) supports the collection of patient-reported outcome measures after discharge and has entered pilot clinical evaluation. The Clinician Virtual Assistant (cVA) enables interaction with stroke care quality data and dashboards and has undergone initial usability testing. The final versions of the pVA and cVA have been designed and are currently being implemented by the technical partners and will be released for testing and evaluation in the Spring of 2026.
Predictive modeling capabilities were developed, including a fully functional Data Science Workbench integrating visualization and predictive tools, and machine learning models for predicting 90-day stroke outcomes and post-stroke conditions, all enhanced by explainable AI features to support clinical trust and transparency. The predictive tool is currently being evaluated with clinicians with synthetic data and could be made available to RES-Q users once MDR approval has been obtained.
A semantic harmonization layer has been developed, incorporating FAIRification infrastructure, a Data Catalog, and ontology mapping through knowledge graphs. The data model has been aligned with standardized clinical terminologies, ensuring interoperability, consistent data representation, and facilitating data integration and reuse across systems, e.g. other registries. Given appropriate approval, future studies could pool data from multiple registries for e.g. analysis of care quality data, predictions of outcomes.
All developments were accompanied by continuous legal and ethical assessment, including data protection impact assessments, data sharing agreements and analysis of regulatory requirements relevant to data processing, artificial intelligence and medical devices.
Figure 2: The RES-Q+ consortium at the M30 project meeting in Murcia, 2025.
RES-Q+ advances the state of the art by enabling automated, multilingual extraction of stroke care quality data from unstructured clinical documents and by supporting the ongoing integration of these data into a registry infrastructure. The project operationalised the concept of discharge report standardisation by delivering a functional tool Discharge Report Composer that generates standardised reports directly from registry data. This represents a shift from conceptual standards towards practical, scalable implementation.
The ongoing development of virtual assistants introduces new methods for collecting patient-reported outcomes and supporting clinicians in analysing stroke care quality data. In addition, predictive models and data imputation methods are being developed to explore relationships between care quality and outcomes.
The ongoing development of virtual assistants introduces new methods for collecting patient-reported outcomes and supporting clinicians in analysing stroke care quality data. In addition, predictive models and data imputation methods are being developed to explore relationships between care quality and outcomes.