A de-identified inpatient clinical pathway dataset integrating ICD-10 diagnoses, ICD-9-CM procedures, Indonesian Case-Based Groups (INA-CBG), and length of stay for healthcare analytics

Background Healthcare administrative databases have become indispensable resources for healthcare analytics, hospital management, clinical pathway analysis, health services research and artificial intelligence. Modern hospital information systems routinely collect standardized diagnoses, procedures, reimbursement classifications, and hospitalization outcomes in everyday clinical practice. Despite their importance, publicly accessible datasets integrating diagnosis codes, procedure codes, case-mix classifications, and hospitalization outcomes remain scarce, particularly in low- and middle-income countries (LMICs). The absence of comprehensive administrative datasets limits reproducible research in clinical pathway modelling, healthcare operations, reimbursement analysis, and machine learning. To address this gap, we developed and publicly released a de-identified inpatient clinical pathway dataset derived from routine hospital administrative records processed through the Indonesian Case-Based Groups (INA-CBG) reimbursement system. Methods The dataset was constructed from finalized inpatient hospitalization records collected between November 2016 and April 2026. Clinical diagnoses were represented using the International Classification of Diseases, Tenth Revision (ICD-10), and clinical procedures were encoded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Each hospitalization episode included one primary diagnosis, up to 11 secondary diagnoses, one primary procedure, up to 16 secondary procedures, one INA-CBG classification code, one INA-CBG description, and one length-of-stay (LOS) variable. Data extraction, transformation, harmonization, de-identification, quality assessment, and validation were performed using a reproducible Python-based computational workflow. The final repository contained 165,391 hospitalization episodes represented by 32 standardized variables. Conclusions The dataset provides an openly available resource for healthcare analytics, clinical pathway research, process mining, hospital management, reimbursement analysis, predictive modelling, explainable artificial intelligence, and healthcare operational research. The combination of standardized ICD-10 diagnoses, ICD-9-CM procedures, INA-CBG classifications, and hospitalization outcomes enables diverse secondary analyses while maintaining interoperability with existing healthcare information systems. All datasets, metadata, source codes, and supporting documentation are publicly available through Zenodo to facilitate transparency, reproducibility, and future methodological development.