Performance Inflation and Reporting Biases in Respiratory Artificial Intelligence: A Data-Driven Landscape Analysis

This dataset supports the manuscript "Performance Inflation and Reporting Biases in Respiratory Artificial Intelligence: A Data-Driven Landscape Analysis." It contains metadata extracted from 446 peer-reviewed studies detailing the application of artificial intelligence and machine learning to respiratory diagnostics. Files Included: Respiratory_AI_Database_v1.csv: The finalized, cleaned dataset containing variables for Target Condition, Signal Type, Model Architecture, Sample Size, and Best Performance Metric. Ontology_Mapping_Table: The human-in-the-loop standardization matrix used to harmonize highly variable clinical terminology into definitive parent categories. Code Availability: The Python scripts used to analyze this dataset and generate the manuscript's figures are available on GitHub at: https://github.com/Hassan-Jubair/Respiratory_AI_Database.git