A multi-agent system for healthcare data privacy

Research has been conducted on distributed data mining, privacy preserving data mining, and multiagent systems; however, integrating the aforementioned research areas require further investigation. Sensitive data sources, such as healthcare organizations, may be reluctant to permit sensitive data to leave the source or permit mining of the sensitive data. Additionally, distributed data mining on sensitive data spanning multiple organizations poses challenges for data privacy and integration. A multi-agent architecture may be employed to anonymize data at the source before transmission to a central repository for knowledge discovery. The advantages of the multi-agent method include customization based on individual organizational requirements, facilitating data integration, maintaining data privacy, and maintaining the integrity of source data. This research explores applying a multi-agent architecture coupled with privacy preserving techniques to mining distributed healthcare data. Results indicate a multi-agent architecture may facilitate data mining across disparate sensitive data sources while maintaining data and knowledge integrity.