Disclosure risk of individuals: A k-anonymity study on health care data related to Indian population

Many private organizations are reluctant to share the health related information to the researchers fearing loss of privacy of data. The non availability of data has potential negative implications for the advancement of medical science, the development of new pharmaceutical products, better diagnosis of disease and for national and micro-level health planning. So in this context, the need for the development of reliable and robust anonymization techniques for data, especially relating to health care data has become equipped. This paper attempts to illustrate the disclosure risk of individuals' health records related to Indian population and analyze the need for the development of suitable mechanisms to protect privacy of individuals. The data we have used for our evaluation purposes are made available to us by the nodal agency International Institute for Population Sciences (UPS), Mumbai. It is the data collected as part of the latest National Family Health Survey conducted in the year 2005, namely NFHS-3. Checking with k-anonymity property, the result shows that some of the individuals are at the risk of disclosure, if the actual table is linked to some other publicly available tables. This paper also attempts to illustrate the need for flexible selection of relevant attributes, especially the Quasi-Identifier (QI) attributes for the wide spread acceptance of knowledge-based systems as per the context.