Secure Integration of Electronic Health Data Using Advanced Machine Learning and Blockchain Technology
نویسندگان
1 Department of Medicine, Sirjan School of Medical Sciences, Sirjan, Iran
2 Department of Occupational Health Engineering, Sirjan School of Medical Sciences, Sirjan, Iran.
3 Department of Health Information Technology, Sirjan School of Medical Sciences, Sirjan, Iran.
doi
10.30476/jhmi.2025.108377.1309چکیده
Introduction: Data integration and privacy preservation in electronic health records (EHRs)remain major challenges. This study combines advanced machine learning and blockchain toimprove integration and security.Methods: Using a synthetic multicenter EHR dataset (patient records, visits, diagnoses,medications, observations, procedures), we evaluated an Irregular Fuzzy Cellular Automata(IFCA) model—which incorporates fuzzy-logic rules—against XGBoost and LightGBM.Preprocessing included complete anonymization and 98.5% missing-value imputation.Machine learning addressed data integration, inconsistency resolution, and classification;HL7-FHIR–like formats and a Hyperledger Fabric consortium blockchain evaluated securedata exchange and access control. Analyses used Python 3.10 and R 4.2.Results: Machine learning (data integrity & classification): IFCA achieved 92% accuracy(F1=0.90, AUC-ROC=0.92), outperforming XGBoost (89%) and LightGBM (90%); ANOVAindicated statistically significant differences (P<0.05). Blockchain & interoperability (security& exchange): data-exchange success was 94%, combined privacy/security score 95%, with92% simulated attack prevention.Conclusion: The combined approach shows promise for EHR integration and privacypreservation. Validation on real multisite EHR data is recommended to confirmgeneralizability.
