Health data integration with Secured Record Linkage: A practical solution for Bangladesh and other developing countries

摘要

Knowledge discovery from various health data repositories requires the incorporation of healthcare data from diversified sources. Maintaining record linkage during the integration of medical data is an important research issue. Researchers have given different solutions to this problem that are applicable for developed countries where electronic health record of patients are maintained with identifiers like social security number (SSN), universal patient identifier (UPI), health insurance number, etc. These solutions cannot be used correctly for record linkage of health data of developing countries because of missing data, ambiguity in patient identification, and high amount of noise in patient information. Also, identifiable health data in electronic health repositories may produce a significant risk to patient privacy and also make the health information systems security vulnerable to hackers. In this paper, we have analyzed the practical problems of collecting and integrating healthcare data in Bangladesh for developing national health data warehouse. We have proposed a privacy preserved secured record linkage architecture that can support constrained health data of developing countries such as Bangladesh. Our technique can anonymize identifiable private data of the patients while maintaining record linkage in integrated health repositories to facilitate knowledge discovery process. Experimental results show that our proposed method successfully linked records with acceptable accuracy for noisy data in the absence of any standard ID like SSN.