The integration of artificial intelligence (AI) into healthcare is transforming clinical decision-making, patient outcomes, and workflows. AI inference, applying trained models to new data, is central to this evolution, with cloud-based infrastructures enabling scalable AI deployment. The Open Medical Inference (OMI) platform democratizes AI access through open protocols and standardized data formats for seamless, interoperable healthcare data exchange. By integrating standards like FHIR and DICOMweb, OMI ensures interoperability between healthcare institutions and AI services while fostering ethical AI use through a governance framework addressing privacy, transparency, and fairness.OMI's implementation is structured into work packages, each addressing technical and ethical aspects. These include expanding the Medical Informatics Initiative (MII) Core Dataset for medical imaging, developing infrastructure for AI inference, and creating an open-source DICOMweb adapter for legacy systems. Standardized data formats ensure interoperability, while the AI Governance Framework promotes trust and responsible AI use.The project aims to establish an interoperable AI network across healthcare institutions, connecting existing infrastructures and AI services to enhance clinical outcomes. · OMI develops open protocols and standardized data formats for seamless healthcare data exchange.. · Integration with FHIR and DICOMweb ensures interoperability between healthcare systems and AI services.. · A governance framework addresses privacy, transparency, and fairness in AI usage.. · Work packages focus on expanding datasets, creating infrastructure, and enabling legacy system integration.. · The project aims to create a scalable, secure, and interoperable AI network in healthcare.. · Pelka O, Sigle S, Werner P et al. Democratizing AI in Healthcare with Open Medical Inference (OMI): Protocols, Data Exchange, and AI Integration. Rofo 2026; 198: 173-184.

BackgroundFast Healthcare Interoperability Resources (FHIR) is a widely used standard for storing and exchanging health care data. At the same time, image-based artificial intelligence (AI) models for quantifying relevant body structures and organs from routine computed tomography (CT)/magnetic resonance imaging scans have emerged. The missing link, simultaneously a needed step in advancing personalized medicine, is the incorporation of measurements delivered by AI models into an interoperable and standardized format. Incorporating image-based measurements and biomarkers into FHIR profiles can standardize data exchange, enabling timely, personalized treatment decisions and improving the precision and efficiency of patient care. ObjectiveThis study aims to present the synergistic incorporation of CT-derived body organ and composition measurements with FHIR, delineating an initial paradigm for storing image-based biomarkers. MethodsThis study integrated the results of the Body and Organ Analysis (BOA) model into FHIR profiles to enhance the interoperability of image-based biomarkers in radiology. The BOA model was selected as an exemplary AI model due to its ability to provide detailed body composition and organ measurements from CT scans. The FHIR profiles were developed based on 2 primary observation types: Body Composition Analysis (BCA Observation) for quantitative body composition metrics and Body Structure Observation for organ measurements. These profiles were structured to interoperate with a specially designed Diagnostic Report profile, which references the associated Imaging Study, ensuring a standardized linkage between image data and derived biomarkers. To ensure interoperability, all labels were mapped to SNOMED CT (Systematized Nomenclature of Medicine – Clinical Terms) or RadLex terminologies using specific value sets. The profiles were developed using FHIR Shorthand (FSH) and SUSHI, enabling efficient definition and implementation guide generation, ensuring consistency and maintainability. ResultsIn this study, 4 BOA profiles, namely, Body Composition Analysis Observation, Body Structure Volume Observation, Diagnostic Report, and Imaging Study, have been presented. These FHIR profiles, which cover 104 anatomical landmarks, 8 body regions, and 8 tissues, enable the interoperable usage of the results of AI segmentation models, providing a direct link between image studies, series, and measurements. ConclusionsThe BOA profiles provide a foundational framework for integrating AI-derived imaging biomarkers into FHIR, bridging the gap between advanced imaging analytics and standardized health care data exchange. By enabling structured, interoperable representation of body composition and organ measurements, these profiles facilitate seamless integration into clinical and research workflows, supporting improved data accessibility and interoperability. Their adaptability allows for extension to other imaging modalities and AI models, fostering a more standardized and scalable approach to using imaging biomarkers in precision medicine. This work represents a step toward enhancing the integration of AI-driven insights into digital health ecosystems, ultimately contributing to more data-driven, personalized, and efficient patient care.

Background Dementia is a widespread syndrome that currently affects more than 55 million people worldwide. Digital screening instruments are one way to increase diagnosis rates. Developing an app for older adults presents several challenges, both technical and social. In order to make the app user-friendly, feedback from potential future end users is crucial during this development process. Objective This study aimed to establish a user-centered design process for the development of digiDEM-SCREEN, a user-friendly app to support early identification of persons with slight symptoms of dementia. Methods This research used qualitative and quantitative methods and involved 3 key stakeholder groups: the digiDEM research team, the software development team, and the target user group (older adults ≥65 years with and without cognitive impairments). The development of the screening app was based on an already existing and scientifically analyzed screening test (Self-Administered Tasks Uncovering Risk of Neurodegeneration; SATURN). An initial prototype was developed based on the recommendations for mobile health apps and the teams’ experiences. The prototype was tested in several iterations by various end users and continuously improved. The app’s usability was evaluated using the System Usability Scale (SUS), and verbal feedback by the end users was obtained using the think-aloud method. Results The translation process during test development took linguistic and cultural aspects into account. The texts were also adapted to the German-speaking context. Additional instructions were developed and supplemented. The test was administered using different randomization options to minimize learning effects. digiDEM-SCREEN was developed as a tablet and smartphone app. In the first focus group discussion, the developers identified and corrected the most significant criticism in the next version. Based on the iterative improvement process, only minor issues needed to be addressed after the final focus group discussion. The SUS score increased with each version (score of 72.5 for V1 vs 82.4 for V2), while the verbal feedback from end users also improved. Conclusions The development of digiDEM-SCREEN serves as an excellent example of the importance of involving experts and potential end users in the design and development process of health apps. Close collaboration with end users leads to products that not only meet current standards but also address the actual needs and expectations of users. This is also a crucial step toward promoting broader adoption of such digital tools. This research highlights the significance of a user-centered design approach, allowing content, text, and design to be optimally tailored to the needs of the target audience. From these findings, it can be concluded that future projects in the field of health apps would also benefit from a similar approach.

BACKGROUND:The current gap between the availability of routine imaging data and its provisioning for medical research hinders the utilization of radiological information for secondary purposes. To address this, the German Medical Informatics Initiative (MII) has established frameworks for harmonizing and integrating clinical data across institutions, including the integration of imaging data into research repositories, which can be expanded to routine imaging data. OBJECTIVES:This project aims to address this gap by developing a large-scale data processing pipeline to extract, convert, and pseudonymize DICOM (Digital Imaging and Communications in Medicine) metadata into "ImagingStudy" Fast Healthcare Interoperability Resources (FHIR) and integrate them into research repositories for secondary use. METHODS:The data processing pipeline was developed, implemented, and tested at the Data Integration Center of the University Hospital Erlangen. It leverages existing open-source solutions and integrates seamlessly into the hospital's research IT infrastructure. The pipeline automates the extraction, conversion, and pseudonymization processes, ensuring compliance with both local and MII data protection standards. A large-scale evaluation was conducted using the imaging studies acquired by two departments at University Hospital Erlangen within 1 year. Attributes such as modality, examined body region, laterality, and the number of series and instances were analyzed to assess the quality and availability of the metadata. RESULTS:Once established, the pipeline processed a substantial dataset comprising over 150,000 DICOM studies within an operational period of 26 days. Data analysis revealed significant heterogeneity and incompleteness in certain attributes, particularly the DICOM tag "Body Part Examined." Despite these challenges, the pipeline successfully generated valid and standardized FHIR, providing a robust basis for future research. CONCLUSION:We demonstrated the setup and test of a large-scale end-to-end data processing pipeline that transforms DICOM imaging metadata directly from clinical routine into the Health Level 7-FHIR format, pseudonymizes the resources, and stores them in an FHIR server. We showcased that the derived FHIRs offer numerous research opportunities, for example, feasibility assessments within Bavarian and Germany-wide research infrastructures. Insights from this study highlight the need to extend the "ImagingStudy" FHIR with additional attributes and refine their use within the German MII.

This single-center retrospective observational study accesses potential differences between adult patients who were admitted to the University Hospital Erlangen between March 2021 and December 2023 (hospital cohort) and adult patients who have given consent to use their documented data for research purposes (broad consent cohort). Demographic and clinical data (ICD-10 diagnoses) were extracted from the university hospitalt's FHIR research data repository and analyzed in pseudonymized form. The two cohorts comprise 98,564 and 1,678 patients respectively and were compared concerning representativity of the BC cohort. The results suggest that the ongoing stepwise rollout of the consent obtainment process creates biases in clinical and demographic characteristics. For as long as these biases persist, we suggest researchers to prefer federated over centralized approaches to data analysis, where broad consent is not required and the analyses can be based on the total hospital cohort.

BACKGROUND:Recruiting patients for the participation in studies is a complex and time-consuming task. OBJECTIVE:The aim was to recruit patients via broad consent to participate in a questionnaire as part of the PEAK study. METHODS:Some patients agreed on the option of recontacting them for future studies as part of the Broad Consent of the Erlangen University Hospital and the University Hospital Halle (Saale). These patients were contacted by letter asking for their participation in the PEAK study although this had no linkage to their illness. 3,489 patients were contacted in Erlangen und 4,000 patients in Halle (Saale). RESULTS:In response, in Erlangen 580 participants (16,6%) completed the questionnaire fully by giving their perspective regarding AI in healthcare. In Halle, 451 participants (11%) completed the questionnaire in full. DISCUSSION:The awareness of the problem of deceased patients has increased, so that a comparison with the population register was made during the course of the process. Furthermore, weighting up between contacting patients for study purposes and the potential withdrawal of consent for recontacting patients in the context of the Broad Consent next to fatigue with regard to participation in studies remains controversial even though there were hardly any significant numbers of withdrawals of consent. CONCLUSION:The utilization of the Broad Consent beyond the mainly intended use for retrospective secondary data use or disease-related study invitation instead now as an instrument for general study recruitment led to a successful recruitment of patients to respond the PEAK patient questionnaire. Contrary to the original fear, only a minimum of patients withdrew their authorization for recontacting. Several learnings, such as the need for a comprehensive vital checking prior the sending, could be derived from this attempt.

BACKGROUND:The accumulation of Real-World Data (RWD) from Electronic Health Records (EHRs) and registries offers substantial potential for generating Real-World Evidence (RWE). However, the ability to generate robust evidence from real-world data hinges on its quality. This is especially critical when heterogeneous data is first transformed into standardized, research-ready data models. OBJECTIVE:This study presents an approach for assessing data completeness through a pipeline for extracting and transforming oncological RWD. METHODS:We introduce a technical solution that enables the assessment of data completeness across three data transformation stages, beginning with the initial data source and extending through Health Level 7 (HL7) Fast Healthcare Interoperability Resources (FHIR) to CSV. RESULTS:Using Trino, a distributed SQL engine, we evaluate data completeness at the three transformation stages by comparing cancer diagnosis counts. The modular pipeline design, compatible with various data sources, allows for error detection in ETL processes. CONCLUSION:Future work will expand the system to address additional data quality dimensions, such as correctness and plausibility, improving the overall robustness of data analytics in federated environments.

Background: Dementia is a growing global health challenge with significant economic and social implications. Underdiagnosis of dementia is prevalent due to a lack of knowledge and understanding among the general population. Enhancing dementia literacy through improved health information-seeking behavior is crucial for the self-determined management of the disease by those affected. Understanding the relationship between dementia literacy, health information-seeking behavior, and the use of various information sources among individuals with cognitive impairment is of high importance in this context. Objective: The aim of this study was to analyze the relevance of different sources of health information from the perspective of people with cognitive impairment, while also evaluating differences based on age, gender, and disease progression. Methods: This study is part of the ongoing project "Digital Dementia Registry Bavaria - digiDEM Bayern." The Digital Dementia Registry Bavaria is a multicenter, prospective, longitudinal register study in Bavaria, Germany. People with cognitive impairment rated several information sources by using Likert scales with the values unimportant (1) to very important (5). Data were analyzed descriptively, and multiple 2-sample, 2-tailed t tests were used to evaluate differences by cognitive status and gender and using multiple one-way ANOVA to evaluate differences by age group. Results: Data of 924 people with cognitive impairment (531 with dementia, 393 with mild cognitive impairment) were evaluated. The most relevant health information sources were "Personal visit to a medical professional" (mean 3.9, SD 1.1) and "Family / Friends" (mean 3.9, SD 1.2). "Internet" was 1 of the 2 lowest-rated information sources by people with cognitive impairment (mean 1.6, SD 1.1), with nearly three-quarters (684/924, 74%) of the participants rating the source as unimportant. The age-specific analyses showed significant differences for the sources "Internet" (F2,921=61.23; P<.001), "Courses / Lectures" (F2,921=18.88; P<.001), and "Family / Friends" (F2,921=6.27; P=.002) for the 3 defined age groups. There were several significant differences between people with mild cognitive impairment and dementia whereby the first group evaluated most sources higher, such as "Internet" (mean difference=0.6; t640=7.52; P<.001). The only sources rated higher by the dementia group were "TV / Radio" and "Family / Friends," with none of them showing significant differences. Gender-specific analyses showed women with cognitive impairment valuing every evaluated source higher than men apart from "Internet" (mean difference=0.4; t685=4.97; P<.001). Conclusions: To enhance health and dementia literacy, the best way to communicate health information to people with cognitive impairment is through interpersonal contact with medical professionals and their friends and family. Slight changes in valuation should be considered as the medical condition progresses, along with variations by age and gender. In particular, the evaluation and use of the internet are dependent on these factors. Further research is needed to capture potential changes in the valuation of the internet as a health information source.

The use of electronic health records for clinical research offers access to large-scale real-world data, but it requires the accurate transformation of data across clinical data repositories. In this study, we evaluate the data quality and completeness in three repositories (DWH, FHIR, and TriNetX) at Erlangen University Hospital. Key data elements (diagnosis, procedure, and laboratory codes) were analyzed, alongside a specific research question. Our results show good overall consistency, but discrepancies arise due to differences in code systems, data filtering, and the mapping process. These findings highlight the importance of critically assessing data provenance and the transformation processes when conducting multicenter research. Understanding the strengths and limitations of each repository is essential for ensuring high-quality research outcomes.

IntroductionDementia is one of the most relevant widespread diseases, with a prevalence of currently 55 million people with dementia worldwide. However, about 60–75% of people with dementia have not yet received a formal diagnosis. Asymptomatic screening of cognitive impairments using neuropsychiatric tests has been proven to efficiently enhance diagnosis rates. Digital screening tools, in particular, provide the advantage of being accessible without spatial or time restrictions. The study aims to validate a digital cognitive screening test (digiDEM-SCREEN) as an app in the German language.Methods and analysisThis is a multicentre study in Bavaria. Participants are people with mild cognitive impairment, people with dementia in an early stage and cognitively healthy people. Recruitment will take place in specialised diagnostic facilities (memory outpatient clinics). 135 participants are aimed based on a power analysis. Sociodemographic data, diagnosis and results of neuropsychiatric tests (Consortium to Establish a Registry for Alzheimer’s Disease, Montreal Cognitive Assessment, digiDEM-SCREEN) will be collected at one point per person via electronic data capturing. The sensitivity, specificity and corresponding cut-off values will be determined based on receiver-operating-characteristic curves. The correlation of the digiDEM-SCREEN test with existing cognitive screening/testing procedures will be analysed.Ethics and disseminationThe study obtained ethical approval from the Ethics Committee of the Julius-Maximilians-Universität of Würzburg (JMU) (application number: 177/23-sc). The test will give feedback about the current cognitive status and possible cognitive impairments that should lead to the users seeking further diagnostic measures by medical professionals. It will be accessible free of charge in established app stores. The results of the validation study will be published in peer-reviewed journals.

INTRODUCTION:Data-driven analysis of clinical databases is an efficient method for clinical knowledge generation, which is especially suitable when exceptional ethical and practical restrictions apply, such as in pediatrics. In the multi-center PEDREF 2.0 study, we are analyzing children's laboratory test results, diagnoses, and procedures from more than 20 German tertiary care centers to establish pediatric reference intervals. The PEDREF 2.0 study uses the framework of the German Medical Informatics Initiative, but the specific study needs require the development of a customized module for distributed pediatric analyses. METHODS:We developed the Pediatric Distributed Analysis, Anonymization, and Aggregation Module (PED-DATA), which is a containerized application that we deployed to all participating centers. PED-DATA transforms the input datasets to a harmonized internal representation and enables their decentralized analysis in compliance with data protection rules, resulting in an anonymous output dataset that is transferred for central analysis. RESULTS:In a preliminary analysis of data from 15 centers, we analyzed 52,807,236 laboratory test results from 753,774 different patients (323,943 to 4,338,317 test results per laboratory test), enabling us to establish pediatric reference intervals with previously unmatched precision. CONCLUSION:PED-DATA facilitates the implementation of pediatric data-driven multicenter studies in a decentralized and privacy-respecting manner, and its use throughout German University Hospitals in the PEDREF 2.0 study demonstrates its usefulness in a real-world use case.

Introduction: The project “digiDEM Bayern” aims to set up a registry with long-term follow-up data on people with dementia and their family caregivers. For that purpose an Electronic Data Capture (EDC) system linked with a Participant Management (PM) system has been established. This study evaluates the acceptance and usability of the IT tools supporting all data management processes in order to further improve the system and associated processes. Methods: For this purpose we collected the key numbers of the registry, and used the System Usability Scale (SUS) to evaluate the interactions of the data management systems in a wide area. Results: Thirty-six research partners (RP) and six study team (ST) members completed the anonymous online survey. The EDC system overall reached an average SUS score of 73.42 and the PM system of 77.92. Discussion: The two systems fulfil their required task and, therefore, simplify the work of the RP in the data collection process and of the ST during the data quality checks. Conclusion: Integrating the used systems is therefore recommended for registry studies in other medical areas.

BackgroundClinical trials (CTs) are crucial for medical research; however, they frequently fall short of the requisite number of participants who meet all eligibility criteria (EC). A clinical trial recruitment support system (CTRSS) is developed to help identify potential participants by performing a search on a specific data pool. The accuracy of the search results is directly related to the quality of the data used for comparison. Data accessibility can present challenges, making it crucial to identify the necessary data for a CTRSS to query. Prior research has examined the data elements frequently used in CT EC but has not evaluated which criteria are actually used to search for participants. Although all EC must be met to enroll a person in a CT, not all criteria have the same importance when searching for potential participants in an existing data pool, such as an electronic health record, because some of the criteria are only relevant at the time of enrollment. ObjectiveIn this study, we investigated which groups of data elements are relevant in practice for finding suitable participants and whether there are typical elements that are not relevant and can therefore be omitted. MethodsWe asked trial experts and CTRSS developers to first categorize the EC of their CTs according to data element groups and then to classify them into 1 of 3 categories: necessary, complementary, and irrelevant. In addition, the experts assessed whether a criterion was documented (on paper or digitally) or whether it was information known only to the treating physicians or patients. ResultsWe reviewed 82 CTs with 1132 unique EC. Of these 1132 EC, 350 (30.9%) were considered necessary, 224 (19.8%) complementary, and 341 (30.1%) total irrelevant. To identify the most relevant data elements, we introduced the data element relevance index (DERI). This describes the percentage of studies in which the corresponding data element occurs and is also classified as necessary or supplementary. We found that the query of “diagnosis” was relevant for finding participants in 79 (96.3%) of the CTs. This group was followed by “date of birth/age” with a DERI of 85.4% (n=70) and “procedure” with a DERI of 35.4% (n=29). ConclusionsThe distribution of data element groups in CTs has been heterogeneously described in previous works. Therefore, we recommend identifying the percentage of CTs in which data element groups can be found as a more reliable way to determine the relevance of EC. Only necessary and complementary criteria should be included in this DERI.

BackgroundCurrently, there is no curative treatment for dementia. The implementation of preventive measures is of great importance. Therefore, it is necessary to identify and address individual and modifiable risk factors. Social isolation, defined through social networks, is a factor that may influence the onset and progression of the disease. The networks of older people are mostly composed of either family or friends. The aim of this study is to examine the influence of social isolation and network composition on cognition over the course of 12 months in people with cognitive impairment.MethodsData basis is the multicentre, prospective, longitudinal register study 'Digital Dementia Registery Bavaria-digiDEM Bayern'. The degree of social isolation was assessed using the Lubben Social Network Scale- Revised (LSNS-R) and the degree of cognitive impairment using the Mini Mental State Examination (MMSE), conducted at baseline and after 12 months. Data were analysed using pre-post ANCOVA, adjusted for baseline MMSE, age, gender, education, living situation and Barthel-Index.Results106 subjects (78.9 ± 8.2 years; 66% female) were included in the analysis. The mean MMSE score at baseline was 24.3 (SD = 3.6). Within the friendship subscore, risk for social isolation was highly prevalent (42.5%). Though, there was no difference between individuals with higher/ lower risk of social isolation within the friendship-network after adjusting for common risk factors in cognitive decline over time, F (1,98) = .046, p = .831, partial η2 = .000.ConclusionThe results of this study showed that the risk of social isolation from friends is very high among people with cognitive impairment. However, social isolation does not appear to have a bearing influence on the course of cognition. Nevertheless, it is important for people with cognitive impairment to promote and maintain close social contacts with friends.

Real-world data (RWD) from sources like administrative claims, electronic health records, and cancer registries offer insights into patient populations beyond the tightly regulated environment of randomized controlled trials. To leverage this and to advance cancer research, six university hospitals in Bavaria have established a joint research IT infrastructure. This article aims to outline the design, implementation, and deployment of a modular data transformation pipeline that transforms oncological RWD into HL7 (Health Level 7) FHIR (Fast Healthcare Interoperability Resources) format and then into a tabular format in preparation for a federated analysis (FA) across the six BZKF university hospitals. To harness RWD effectively, we designed a pipeline to convert the oncological basic dataset (oBDS) into HL7 FHIR format and prepare it for federated analysis. The pipeline handles diverse IT infrastructures and systems while maintaining privacy by keeping data decentralized for analysis. To assess the functionality and validity of our implementation, we defined a cohort to address two specific medical research questions. We evaluated our findings by comparing the results of the FA with reports from the Bavarian Cancer Registry and the original data from local tumor documentation systems. We conducted a federated analysis of 17,885 cancer cases from 2021/2022. Breast cancer was the most common diagnosis at three sites, prostate cancer ranked in the top two at four sites, and malignant melanoma was notably prevalent. Gender-specific trends showed larynx and esophagus cancers were more common in males, while breast and thyroid cancers were more frequent in females. Discrepancies between the Bavarian Cancer Registry and our data, such as higher rates of malignant melanoma (5 % vs. 11 %) and lower representation of colorectal cancers (13 % vs. 7 %) likely result from differences in the time periods analyzed (2019 vs. 2021/2022) and the scope of data sources used. The Bavarian Cancer Registry reports approximately three times more cancer cases than the six university hospitals alone. The modular pipeline successfully transformed oncological RWD across six hospitals, and the federated approach preserved privacy while enabling comprehensive analysis. Future work will add support for recent oBDS versions, automate data quality checks, and integrate additional clinical data. Our findings highlight the potential of federated health data networks and lay the groundwork for future research that can leverage high-quality RWD, aiming to contribute valuable knowledge to the field of cancer research.

In the light of big data driven clinical research, fair access to real world clinical health data enables evidence to improve patient care. Germany's healthcare system provides an abundant data resource but unique challenges due to its federated nature, heterogeneity and high data-protection standards. The Medical Informatics Initiative (MII) developed concepts that are being implemented in the German Portal for Medical Research Data (FDPG) to grant access to distributed data-sources across state borders. The portal currently provides access to more than 10 million patient resources containing hundreds of millions of laboratory parameters, diagnostic reports, administered medications, procedures and specimens. Upcoming datasets include among others oncological data, molecular analysis results and microbiological findings. Here, we describe the philosophy, implementation and experience behind the framework: standardized access processes, interoperable fair data, software for in depth feasibility requests, tools to support researchers and hospital stakeholders alike as well as transparency measures to provide data use information for patients. Challenges remain to improve data quality and automatization of technical and organizational processes.

Background: The objective of this IRB-approved retrospective monocentric study was to identify risk factors for mortality after surgery for congenital heart defects (CHDs) in pediatric patients using machine learning (ML). CHD belongs to the most common congenital malformations, and remains the leading mortality cause from birth defects. Methods: The most recent available hospital encounter for each patient with an age <18 years hospitalized for CHD-related cardiac surgery between the years 2011 and 2020 was included in this study. The cohort consisted of 1302 eligible patients (mean age [SD]: 402.92 [±562.31] days), who were categorized into four disease groups. A random survival forest (RSF) and the ‘eXtreme Gradient Boosting’ algorithm (XGB) were applied to model mortality (incidence: 5.6% [n = 73 events]). All models were then applied to predict the outcome in an independent holdout test dataset (40% of the cohort). Results: RSF and XGB achieved average C-indices of 0.85 (±0.01) and 0.79 (±0.03), respectively. Feature importance was assessed with ‘SHapley Additive exPlanations’ (SHAP) and ‘Time-dependent explanations of machine learning survival models’ (SurvSHAP(t)), both of which revealed high importance of the maximum values of serum creatinine observed within 72 h post-surgery for both ML methods. Conclusions: ML methods, along with model explainability tools, can reveal interesting insights into mortality risk after surgery for CHD. The proposed analytical workflow can serve as a blueprint for translating the analysis into a federated setting that builds upon the infrastructure of the German Medical Informatics Initiative.

Clinical trials (CTs) are foundational to the advancement of evidence-based medicine and recruiting a sufficient number of participants is one of the crucial steps to their successful conduct. Yet, poor recruitment remains the most frequent reason for premature discontinuation or costly extension of clinical trials. We designed and implemented a novel, open-source software system to support the recruitment process in clinical trials by generating automatic recruitment recommendations. The development is guided by modern, cloud-native design principles and based on Health Level 7 (HL7) Fast Healthcare Interoperability Resources (FHIR) as an interoperability standard with the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) being used as a source of patient data. We evaluated the usability using the system usability scale (SUS) after deploying the application for use by study personnel. The implementation is based on the OMOP CDM as a repository of patient data that is continuously queried for possible trial candidates based on given clinical trial eligibility criteria. A web-based screening list can be used to display the candidates and email notifications about possible new trial participants can be sent automatically. All interactions between services use HL7 FHIR as the communication standard. The system can be installed using standard container technology and supports more sophisticated deployments on Kubernetes clusters. End-users (n = 19) rated the system with a SUS score of 79.9/100. We contribute a novel, open-source implementation to support the patient recruitment process in clinical trials that can be deployed using state-of-the art technologies. According to the SUS score, the system provides good usability.

The digital health progress hubs pilot the extensibility of the concepts and solutions of the Medical Informatics Initiative to improve regional healthcare and research. The six funded projects address different diseases, areas in regional healthcare, and methods of cross-institutional data linking and use. Despite the diversity of the scenarios and regional conditions, the technical, regulatory, and organizational challenges and barriers that the progress hubs encounter in the actual implementation of the solutions are often similar. This results in some common approaches to solutions, but also in political demands that go beyond the Health Data Utilization Act, which is considered a welcome improvement by the progress hubs.In this article, we present the digital progress hubs and discuss achievements, challenges, and approaches to solutions that enable the shared use of data from university hospitals and non-academic institutions in the healthcare system and can make a sustainable contribution to improving medical care and research.

Die vom Bundesministerium für Bildung und Forschung (BMBF) 2016–2027 geförderte Medizininformatik-Initiative (MII) schafft erfolgreich Grundlagen für die datenbasierte Medizin in Deutschland. Zur Stärkung der Lehre, Aus- und Fortbildung im Bereich der Medizininformatik und zur Kompetenzverbesserung in den medizinischen Datenwissenschaften wurden im Rahmen dieser Förderung 51 neue Professuren, 21 wissenschaftliche Nachwuchsgruppen und verschiedene neue Studiengänge eingerichtet. Eine die gesamte Universitätsmedizin und ihre Partner umfassende gemeinsame dezentral föderierte Forschungsdateninfrastruktur wurde in Gestalt der Datenintegrationszentren (DIZ) an allen Standorten und dem Deutschen Forschungsdatenportal für Gesundheit (FDPG) als zentralem Zugangspunkt geschaffen. Für die Sekundärnutzung von Behandlungsdaten wurde ein modularer Kerndatensatz (KDS) definiert und unter konsequenter Nutzung internationaler Standards (z. B. FHIR, SNOMED CT, LOINC) implementiert. Als Rechtsgrundlage wurde eine behördlich genehmigte bundesweite breite Einwilligung (Broad Consent) eingeführt. Erste Datenausleitungen und Datennutzungsprojekte sind durchgeführt worden, eingebettet in eine übergeordnete Nutzungsordnung und standardisierte vertragliche Regelungen. Die Weiterentwicklung der MII-Gesundheitsforschungsdateninfrastrukturen im kooperativen Rahmen des Netzwerks Universitätsmedizin (NUM) bietet einen hervorragenden Ausgangspunkt für einen deutschen Beitrag zum kommenden Europäischen Gesundheitsdatenraum (EHDS), der Chancen für den Medizinforschungsstandort Deutschland eröffnet.