Healthcare is the dominant item in modern economies, and rapidly growing as the population ages. Current health systems are oriented towards acute illness in hospitals -- there is no current infrastructure that can handle the coming scale of chronic illness in homes. New information technologies can support viable healthcare infrastructure. Health monitors can measure the status of people in their homes and automatically route them to the appropriate provider. Such an infrastructure can be supported on a national scale, by providing personalized interactions via computer that generate detailed databases of patient records. Local interactions track progress of chronic conditions, while global analyses of individual records discover similar cases to guide individual healthcare.

My laboratory is building health monitor prototypes. These systems perform adaptive question asking, using quality of life questionnaires to ask targeted questions to individual patients, customized to their current conditions. The patient answers build a structured vector for each
individual, precisely describing their current health status. These vectors can then be statistically clustered, to determine cohorts of similar patients requiring similar treatments. This talk describes concepts of Internet Health Monitors and preliminary experiments, along with plans for large-scale clinical trials within my new department.