Datasets for: Trust Gap in Clinical Artificial Intelligence: A Meta-Systematic Review (MSR) - 2022-2025

Description

This dataset contains the complete analytical materials for the meta-systematic review (MSR) titled "Trust Gap in Clinical Artificial Intelligence: A Meta-Systematic Review." The study systematically analyzes 130 systematic reviews published between 2022 and 2025 to investigate patterns of ethical engagement, with a specific focus on conceptualizing trust and identifying "ethics-washing" in clinical AI evaluation literature. The dataset supports the study's findings, which reveal a critical trust gap (0% substantive conceptualization of trust) and systematic "ethics-washing" (overall validation score of 21.4%) between claimed and actual ethical engagement. Primary Dataset Includes: Coded Database of 130 Systematic Reviews: The core dataset containing coded variables for each reviewed article. Clinical Dimension: Specialty (e.g., Emergency Medicine, Oncology), application domain, risk level. Methodological Dimension: Review type (e.g., Meta-Analysis, Hybrid AI-assisted), quality assessment (AMSTAR-2 scores). Technical AI Dimension: AI techniques (e.g., Deep Learning, XAI), performance metrics. Ethical Dimension: Dual-coded analysis of ethical engagement (Categorization Claims vs. Content-Based Analysis) for trust, accountability, explainability, bias, and transparency. Dual-Analysis Validation Metrics: Raw data and calculations supporting the quantification of "ethics-washing," including divergence percentages and the Overall Validation Score (OVS) for each ethical concept. CAIEE Framework Operationalization Data: Data and coding schemes used to develop and propose the multidimensional Clinical AI Ethical Engagement (CAIEE) Framework, including indicators for Technical, Interpersonal, Institutional, and Epistemic trust dimensions. Analysis Scripts: Custom Python (using pandas, scikit-learn) and R (using metafor, ggplot2) scripts for quantitative analysis, statistical testing, and visualization generation. Qualitative Codebook: The detailed codebook used for thematic and content analysis, with explicit inclusion/exclusion criteria for operationalizing ethical concepts. Potential Reuse: This dataset is invaluable for researchers, bioethicists, and policymakers interested in: Conducting secondary analysis on ethical trends in AI and digital health. Studying research methodology, evidence synthesis, and reporting quality in systematic reviews. Developing or validating frameworks for the ethical assessment of AI in healthcare. Understanding the phenomenon of "ethics-washing" in technology literature. Teaching concepts of research integrity, meta-science, and critical appraisal. Keywords: Clinical Artificial Intelligence, Trust, Ethics-Washing, Systematic Review, Meta-Systematic Review, Algorithmic Bias, Explainable AI (XAI), Research Integrity, CAIEE Framework, Dual-Analysis Methodology.

Institutions

• Budapesti Corvinus Egyetem

  Budapest

Categories

Licence