Dataset corresponding to the manuscript "Leveraging an explainable machine learning model for early identification of Acute Kidney Injury: A retrospective study".

The project was carried out through a collaborative effort between The Kokilaben Dhirubhai Ambani Hospital and Medical Research Institute and The Indian Institute of Technology Bombay. This single-center, retrospective observational study was conducted using inpatient data from the Kokilaben Dhirubhai Ambani Hospital and Medical Research Institute, Mumbai, India. The study covered admissions between January 1 and August 31, 2023, and received approval for public data sharing from the Institutional Ethical Committee (IEC–A Code: 37/2021). All inpatient records included in the analysis were fully anonymized to ensure patient confidentiality. The dataset comprised admission details, diagnoses, vitals, comorbidities, clinical findings, with AKI-related biochemical and hematological features selected through literature review and clinician consensus. Rate-of-change features were computed using admission values and the serum creatinine alert point for AKI patients or the discharge value for non-AKI patients.