Data made available toresearch groups investigating everything from cancer clusters to therisk of living near to hazardous waste sites is often restricted,altered or aggregated in order to protect the identity of individualpatients.
But researchers say that these measures often make itimpossible for them to carry out accurate geographical analyses ofpublic health concerns, and may even result in misleading informationbeing used in healthcare decisions.
They suggest that newtechnology which uses software “agents” to explore data could providehealthcare professionals with more accurate and meaningful informationwithout risking patients’ identities being revealed.
Agents areadvanced software programmes that can be set a specific task but thengiven the autonomy to set goals and carry out the operations necessaryto achieve them.
By constructing virtual institutions in whichagents can act, collaborating organisations can make raw data availablefor research without compromising the security of the information.
“Itis becoming increasingly clear that certain measures to protectindividual privacy can destroy the information needed for geographicalanalyses, making it impossible to address many important public healthconcerns,” said Dr Maged Boulos, from the School for Health at theUniversity of Bath.
“Some of the solutions used to preserveconfidentiality, such as centralising information to a single point ina town or aggregating data to cover the whole of an area, either lackthe flexibility healthcare researchers need to get the information theyneed, or else actually obscure the results.
“This degrades theability of public health researchers to identify, for example, the riskof exposure to lead associated with urban highways or clusters ofcancer cases.
“Such widespread concerns can only be addressedusing micro data and access to this often involves lengthy andcumbersome procedures through review boards and committees forapproval, and sometimes it is just not possible.”
Dr Boulos,together with colleagues from the Department of Computer Science at theUniversity of Bath and the University of Iowa, suggests that new agentsoftware may be able to overcome some of these problems.
Softwareagents are programmes that can respond to changes in their environment,generate and attempt to achieve goals, and have the capacity tointeract with other agents and even co-operate.
This means thatagents could be sent to the original data repository in order to carryout the analysis there, and then send back an aggregate report thatdoes not reveal individual identities. “Software agents could provideflexible but controlled access to unmodified confidential data, andreturn only results that do not expose any person-identifiabledetails,” said Dr Boulos.
“The use of software agents is not asimple as it sounds, and also carries with it its own security risks,which must be properly addressed.
“Mechanisms need to beintroduced that, for example, digitally sign and authenticate genuineagents and their transactions, and prevent ‘Trojan horse’-like attacksby fake or rogue agents.
“These mechanisms could include thecreation of virtual institutions to insulate host organisations fromagents and minimise leakage by limiting access to only the necessarydata.”
Cite This Page: