Diabetes Clinical Dataset Derived from Electronic Health Records (EHR)

This dataset contains anonymized clinical records collected retrospectively from 30 healthcare centers located in Istanbul, Turkey. The data were extracted from routinely collected electronic health records and represent real-world clinical measurements obtained during standard medical practice. The complete dataset includes 118,265 clinical records and contains demographic, clinical, and laboratory variables relevant to diabetes assessment and metabolic health evaluation. These variables include patient demographics, blood pressure measurements, anthropometric data, laboratory test results, and family history indicators.The file included in this version, "Diabetes_Raw_Dataset_Example.csv", contains a small illustrative subset of the full dataset and is provided for structural reference and reproducibility testing purposes. The complete raw dataset containing 118,265 records (Diabetes_Raw_Dataset.csv) will be publicly released upon publication of the associated research article as a new version of this Zenodo record.The dataset is provided in raw form. As such, the full dataset may contain duplicate entries, missing values, and measurement inconsistencies that are commonly present in real-world clinical data. No preprocessing, normalization, imputation, outlier removal, feature engineering, or data cleaning procedures were applied prior to sharing.All direct and indirect personal identifiers were removed before release. The dataset therefore contains fully anonymized records and does not allow identification of individual patients.Ethical approval for the original data collection was obtained from the Istanbul Medipol University Non-Interventional Clinical Research Ethics Committee (Decision No: 685, dated 03 September 2020). Subsequent protocol amendments were approved by the same committee on 25 June 2024. Individual informed consent was not obtained because the dataset consists of retrospectively collected and fully anonymized clinical records.This dataset is shared to promote transparency, reproducibility, and methodological development in artificial intelligence and machine learning research applied to healthcare and clinical decision support systems. It enables independent validation studies, benchmarking of analytical methods, and development of predictive models using real-world clinical data.This dataset is intended strictly for research and educational purposes. It must not be used for direct clinical decision making, patient diagnosis, or medical treatment.