Rare Diseases
Conditions
Interventions
Group 1: In the prospective study arm, medical documents submitted as part of routine clinical care are analyzed for newly enrolled patients.
Based on the same documents, two case summaries are create
Sponsors
Institut für Digitale Allgemeinmedizin
Eligibility
Sex/Gender
All
Inclusion criteria
Inclusion criteria: Patients treated at the Center for Rare Diseases (ZSEA) who provide informed consent for the parallel creation of a case summary. For patients under 18 years of age, inclusion is permitted if the legal guardians provide informed consent for participation. Medical documents must be available in German. Vulnerable groups are included in this study. This includes patients under 18 years of age. The inclusion of minors is necessary, as rare diseases frequently affect children, particularly in the case of genetic disorders.
Exclusion criteria
Exclusion criteria: There are no specific exclusion criteria.
Design outcomes
Primary
| Measure | Time frame |
|---|---|
| Primary endpoints: The primary endpoint of the study is the standardized overall quality score (0–100 points) of the generated case report, assigned by an independent, blinded expert panel. The score is calculated as a composite sum score based on four predefined subcriteria (completeness, structural clarity, diagnostic traceability, and clinical relevance). The primary analysis is the comparison of the overall quality score between AI-generated and manually created case reports based on identical source documents. The analysis is conducted as a paired comparison, since both case reports are created for each case. Additional analyses include: – Calculation of interrater reliability of the overall quality score using the intraclass correlation coefficient (ICC) – Comparison of time expenditure between manual and AI-assisted creation – Sensitivity analysis with separate comparison of results in the prospective and retrospective study arms | — |
Secondary
| Measure | Time frame |
|---|---|
| Secondary endpoints: The secondary endpoints of the study are: Processing time per case (in minutes) for manual versus AI-assisted generation of the case report, to quantify a potential reduction in time expenditure. The number of clinically relevant information losses per case report. Information loss is defined as the absence of predefined decision-relevant anamnestic data, findings, or diagnostic results. Interrater reliability of the quality assessment, calculated using the intraclass correlation coefficient (ICC) to determine agreement among members of the expert panel. Comparison of the primary and secondary endpoints between the prospective and retrospective study arms as an exploratory analysis to assess potential context or data effects. | — |
Countries
Germany
Contacts
Public ContactMartin Mücke
Institut für Digitale Allgemeinmedizin
Outcome results
None listed