'[The Care Engine] will look at individual records for evidence across multiple data points to allow it to create a statistically valid conclusion,' Heyl said. 'Whatever conclusions are made, whatever your health status is, [it] gets summarized into a PHR.'

The online component that allows members access to the PHR is a Microsoft .Net-based application that uses Web services to transmit information between the online interface and ActiveHealth's databases, Heyl said. While the PHR data will be stored on a Microsoft SQL Server database, ActiveHealth is moving the CareEngine from SQL to run on an Oracle database to add advanced analytics, he said.

Although the PHR application is built in .Net, Aetna wanted to incorporate some of the patient data from the PHR into various internal Java applications -- such as call center and disease management applications -- that use IBM's WebSphere middleware. 'We've placed an intermediary between the Java and .Net interactions that works on XML to make sure ... we remove any kind of .Net-skewed XML data types or Java-skewed data types,' Heyl said.

While Aetna considered using an enterprise service bus (ESB) as the intermediary to do the transformation needed for the applications to interoperate, it opted to write its own custom code for that job because of concerns that an ESB might not provide the scalability needed for the PHR, he said.

The biggest challenge that Aetna has encountered in building the PHR is managing individual patient identities and the credentials used to authenticate identity, Heyl said. The insurer is using industry standards related to federated identity management along with IBM's Tivoli software and Microsoft's Active Directory for identity management.