Actually, we already have the technology we need. A few examples:
We have CPOEs, computerized physician order entry systems, with which doctors can file prescriptions and give orders for lab work. These can issue alerts on possible adverse drug interactions. Mistakes are reduced, as are time and costs.
We have EHRs, electronic health records, which capture every piece of information about a patient and are accessible to qualified medical personnel online. In an emergency the patient doesn't have to remember drugs he's taking and nobody has to track down a manila folder.
We have CDSS, clinical decision support systems. These advise doctors on the latest best practices in diagnosis and treatment, allowing them to customize what they do for an individual patient. This is controversial terrain. There is obviously a difference between what is known in the aggregate and what is right for an individual. And nobody wants insurance companies to use this information to turn doctors into robots; there is enough policing of physicians in the name of cost-containment today.
Nevertheless, when studies indicate that only about half of all patients get widely accepted and uncontroversial advice from their doctors - such as taking aspirin for heart conditions - it's time we considered something new. Moreover, if we had a national, interoperative, medical information network, we would be able to see patterns in the aggregate data. We could learn, for example, whether a certain test is actually worthwhile in a certain situation. The quality of healthcare would go up and the cost would go down.