Christof von Kalle

Objective: To systematically review and characterize methodological heterogeneity in sepsis case detection using the Medical Information Mart for Intensive Care III (MIMIC-III) and eICU-CRD databases. Materials and Methods: We conducted a PRISMA-guided systematic review of PubMed and Web of Science (publication years 2016-2024). Studies were eligible if they applied the SOFA-based Sepsis-3 definition to MIMIC-III or eICU-CRD, provided at least one explicit label-calculation detail, reported a traceable detection rate, and met further criteria detailed in the PRISMA flow diagram. We classified each study by study type and unit of analysis (patients, admissions, or ICU stays) and extracted methodological details on sepsis case detection across 6 domains. For studies with available source code, we additionally examined code structure and repository dependencies to identify methodological decisions across these domains. Results: Of 396 publications screened, 64 met inclusion criteria and 12 provided source code. Sepsis detection rates ranged from 3.4% to 65.2% in MIMIC-III and from 9.8% to 47.9% in eICU-CRD. Substantial variability persisted among studies using identical cohort definitions within both databases (MIMIC-III: 16.9%-42.2%; eICU-CRD: 13.9%-31.4%). In a generalized linear mixed model, between-study heterogeneity was extreme (overall τ²≈0.79; 95% prediction interval for a new study’s detection rate 5%–63%), and study-level characteristics—including database, study type, infection detection method, code availability, and unit of analysis—explained only a small proportion of this variance (≈9%), with no characteristic reaching statistical significance. The overall proportion of studies reporting methodological details varied by domain: SOFA calculation (53.1%), infection detection methods (42.2%), temporal windows (37.5%), aggregation methods (26.6%), and missing-data handling (17.2%). Source code analysis identified 321 implementation decisions, revealing heterogeneity in baseline SOFA definitions (SOFA = 0 vs dynamic baseline), temporal windows (infection-centered vs ICU-admission-centered) and infection detection methods (antibiotic-culture matching vs APACHE-based diagnosis). Dependencies among several MIMIC-III repositories suggested propagation of implementation decisions across studies. Discussion: Clinically validated sepsis definitions yield substantially different detection rates across studies using identical datasets. Statistical modeling showed that this variability is driven almost entirely by between-study implementation differences rather than sampling error or study characteristics. Conclusion: To improve reproducibility in sepsis research and the robustness of sepsis prediction models, we recommend standardized reporting of sepsis case detection methodology and the publication of version-controlled source code.

Objective: To systematically review and characterize methodological heterogeneity in sepsis case detection using the MIMIC-III and eICU-CRD databases. Materials and Methods: We conducted a PRISMA-guided systematic review of PubMed and Web of Science (publication years 2016–2024). We extracted methodological details on sepsis case detection across six domains: parameter coverage, temporal windows, aggregation methods, missing-data handling, SOFA calculation, and infection detection methods. For studies with available source code, we additionally examined code structure and repository dependencies to identify methodological decisions across these domains. Results: Of 396 publications screened, 64 met the inclusion criteria and 12 provided available source code. Sepsis detection rates ranged from 3.4% to 65.2% in MIMIC-III and from 9.8% to 47.9% in eICU-CRD. Substantial variability persisted among studies using identical cohort definitions within both databases (MIMIC-III: 16.9%–42.2%; eICU-CRD: 13.9%–31.4%). The overall proportion of studies reporting methodological details varied by domain: SOFA calculation (53.1%), infection detection methods (42.2%), temporal windows (37.5%), aggregation methods (26.6%) and missing-data handling (17.2%). Source code analysis identified 321 implementation decisions, revealing heterogeneity in baseline SOFA definitions (SOFA=0 vs dynamic baseline), temporal windows (infection-centered vs ICU-admission-centered) and infection detection methods (antibiotic-culture matching vs APACHE-based diagnosis). Dependencies among several MIMIC-III repositories suggested propagation of implementation decisions across studies. Discussion: Clinically validated sepsis definitions yield substantially different detection rates across studies using identical datasets, indicating heterogeneity in computational implementation. Conclusion: To improve reproducibility in sepsis research and the robustness of sepsis prediction models, we recommend standardized reporting of sepsis case detection methodology and the publication of version-controlled source code.

Zusammenfassung: Patient:innenautonomie gilt als zentrales ethisches und gesundheitspolitisches Leitprinzip moderner Versorgung. Ihre praktische Ausgestaltung gewinnt angesichts der rasch voranschreitenden digitalen Transformation des Gesundheitswesens zunehmend an Bedeutung. Digitale Technologien verändern die Bedingungen, unter denen Autonomie ausgeübt und gesichert werden kann. Sie verändern den Zugang zu Informationen und die Orientierung im Versorgungsgeschehen und haben damit Einfluss auf individuelle Entscheidungsprozesse. Vor dem Hintergrund dieser Entwicklungen erweist sich ein rein individualistisch verstandenes Autonomiekonzept als unzureichend. Soll die Digitalisierung des Gesundheitswesens tatsächlich zur Stärkung der Patient:innenautonomie beitragen, muss über die technische und organisatorische Komponente hinaus Autonomie konzeptionell mitgedacht und strukturell berücksichtigt werden. Daraus ergibt sich ein politisch- und sozialstruktureller Gestaltungsauftrag, der weit über die Einführung einzelner Anwendungen hinausreicht. Die konsequente Berücksichtigung eines erweiterten Autonomieverständnisses stellt hohe Anforderungen an digitale Infrastrukturen und deren systematische Einbettung. Digitale Lösungen müssen verlässlich funktionieren, gesellschaftliche Vielfalt berücksichtigen, Teilhabe ermöglichen und Orientierung fördern. Um informationelle und partizipative Selbstbestimmung abzusichern, ist zudem ein Zusammenspiel zielgerichteter kohärenter Regulierung, politischer Steuerung und sozialsensibler struktureller Bedingungen erforderlich. Ziel des Beitrags ist es, diese Anforderungen differenziert herauszuarbeiten und sie mit Blick auf eine patient:innenorientierte Ausgestaltung digitaler Infrastrukturen als emanzipatorischen Gestaltungsauftrag zu profilieren.

Background: Patients with breast cancer frequently experience significant uncertainty, prompting them to seek detailed, personalized, and reliable medical information to enhance adherence to prescribed treatments, medications, and recommended lifestyle adjustments. Although high-quality information exists within oncology guidelines and patient-oriented resources, the provision of tailored responses to individual patient queries remains challenging, especially for non–English-speaking populations. Objective: This study aims to evaluate the potential of an artificial intelligence–driven chatbot, specifically leveraging ChatGPT (GPT-4; OpenAI) combined with retrieval-augmented generation, to deliver personalized answers to complex breast cancer-related patient questions in German. Methods: We collaborated with one of Germany’s largest breast cancer Patient Representation Groups to collect authentic patient inquiries, receiving a total of 118 questions. After initial screening, we selected 104 medical questions, organized into 7 distinct categories: aftercare, bone health, ductal carcinoma in situ, diagnostics, nutrition and supplements, complementary medicine, and therapy. A customized version of GPT-4 was configured with specific system prompts emphasizing empathetic, evidence-based responses and integrated with a comprehensive database comprising guidelines, recommendations, and patient information materials published by recognized German medical societies. To assess chatbot responses, we used 4 evaluation criteria: comprehensibility (clarity from a patient perspective), correctness (accuracy per current medical guidelines), completeness (inclusion of all relevant aspects), and potential harm (risk of undue patient harm or misinformation). Ratings were conducted using a 5-point Likert scale by a breast cancer expert (correctness, completeness, and potential harm) and patient representatives (comprehensibility). Results: The chatbot provided high-quality responses across multiple dimensions. Of the 499 responses evaluated for comprehensibility, 427 (85.6%) were rated as comprehensible. Among the 104 responses assessed for the remaining dimensions, 91 (87.5%) were rated as correct, 72 (69.2%) as complete, and 93 (89.4%) as nonharmful. Reasons for incomplete answers included omission of reimbursement details, updates from recent therapeutic guidelines, or nuanced recommendations regarding endocrine therapy and aftercare schedules. In addition, 6 (5.8%) of the answers were rated as potentially harmful due to outdated or contextually inappropriate recommendations. The chatbot also performed well in the nutrition and bone health categories despite occasionally incomplete document retrieval. Conclusions: Our findings demonstrate that an artificial intelligence–powered chatbot with GPT-4 and retrieval augmentation can effectively provide personalized, linguistically accessible, and largely accurate information to German-speaking patients with breast cancer. This approach holds considerable promise for improving patient-centered communication, empowering patients to make informed decisions. Nonetheless, observed limitations regarding response completeness and potential harm underscore the critical need for ongoing human oversight. Future research and development should prioritize regularly updated databases, advanced retrieval methods to handle complex document structures, multimodal capabilities, and clearly articulated disclaimers emphasizing the necessity of professional medical consultation. Our evaluation, along with the provided set of realistic patient questions, establishes a benchmark for future development and validation of German-language oncology chatbots.

Background: The COVID-19 pandemic has spurred large-scale, interinstitutional research efforts. To enable these efforts, researchers must agree on data set definitions that not only cover all elements relevant to the respective medical specialty but also are syntactically and semantically interoperable. Therefore, the German Corona Consensus (GECCO) data set was developed as a harmonized, interoperable collection of the most relevant data elements for COVID-19–related patient research. As the GECCO data set is a compact core data set comprising data across all medical fields, the focused research within particular medical domains demands the definition of extension modules that include data elements that are the most relevant to the research performed in those individual medical specialties. Objective: We aimed to (1) specify a workflow for the development of interoperable data set definitions that involves close collaboration between medical experts and information scientists and (2) apply the workflow to develop data set definitions that include data elements that are the most relevant to COVID-19–related patient research regarding immunization, pediatrics, and cardiology. Methods: We developed a workflow to create data set definitions that were (1) content-wise as relevant as possible to a specific field of study and (2) universally usable across computer systems, institutions, and countries (ie, interoperable). We then gathered medical experts from 3 specialties—infectious diseases (with a focus on immunization), pediatrics, and cardiology—to select data elements that were the most relevant to COVID-19–related patient research in the respective specialty. We mapped the data elements to international standardized vocabularies and created data exchange specifications, using Health Level Seven International (HL7) Fast Healthcare Interoperability Resources (FHIR). All steps were performed in close interdisciplinary collaboration with medical domain experts and medical information specialists. Profiles and vocabulary mappings were syntactically and semantically validated in a 2-stage process. Results: We created GECCO extension modules for the immunization, pediatrics, and cardiology domains according to pandemic-related requests. The data elements included in each module were selected, according to the developed consensus-based workflow, by medical experts from these specialties to ensure that the contents aligned with their research needs. We defined data set specifications for 48 immunization, 150 pediatrics, and 52 cardiology data elements that complement the GECCO core data set. We created and published implementation guides, example implementations, and data set annotations for each extension module. Conclusions: The GECCO extension modules, which contain data elements that are the most relevant to COVID-19–related patient research on infectious diseases (with a focus on immunization), pediatrics, and cardiology, were defined in an interdisciplinary, iterative, consensus-based workflow that may serve as a blueprint for developing further data set definitions. The GECCO extension modules provide standardized and harmonized definitions of specialty-related data sets that can help enable interinstitutional and cross-country COVID-19 research in these specialties.