Reprinted from The IEEE Technology Time Machine
What’s Ahead at Hong Kong JUNE 1-3, 2011
Today’s reality in health care is that a lot of records and prescriptions are still written by hand, often barely legibly, and medical histories are often not readily available except to those who compiled them in the first place—assuming the doctor’s office or clinic has not in the meantime suffered a fire, flood, or some other natural disaster. Much of that will change dramatically over the next 10 to 20 years, and yet the change will not be total or universal.
It seems safe to say that by 2020 all medical records and doctor’s orders will be created electronically, at least in the advanced industrial and rapidly industrializing countries. Much more medical information for individuals will be integrated in computer files and readily accessible just about anywhere and at any time, once security access queries are satisfied.
But much of the health delivery system will still be fragmented, and many countries, clinics, and clinical systems will still handle their records in ways that prevent them from being instantly retrievable, regardless of what system a patient is in. It will be at least another decade until medical records are fully integrated and made universally accessible throughout and among all developed countries. This will take even longer among the less well-off nations, where basic health concerns take higher priority.
Two models for electronic record keeping have emerged in recent years. In the more familiar one, a hospital or clinic, a network of health-care providers, or even a national health-care system keeps uniform records that are fully accessible internally in the system. Sensitivities about personal information and data security naturally lead to restrictions on access. Yet these systems have amply proven their worth, both in terms of the efficiency and effectiveness with which medical services are provided and in terms of preventing medical errors.
In Europe, the Scandinavian countries and the Netherlands have pioneered such systems. Sweden uses the same unique personal identifier that is used for tax records. Norway has made strides incorporating distance medicine and telemetry into its system because of its scattered population in the country’s far north. Though the United States lags far behind in this area, organizations that have digital record keeping—notably the United States Department of Veterans Affairs—have markedly improved the quality of health care.
An alternative model, which Microsoft and others have been exploring, developing, and promoting, is client centered. In this kind of system, it is the individual who is ultimately responsible for maintaining an integrated case history, which can be more or less comprehensive. (Some may want to include information, for instance, about a particular massage technique they have found to be helpful, while others may prefer not to include such details. If the patient’s visit to a psychiatrist is to be covered, the insurance company will need to know about it, but does the patient’s employer need to know as well?) One obvious advantage of the patient-managed system is that the individual can take full control of all issues concerning confidentiality and security. But it also puts more responsibility on the patient than some care to (or can) assume.
Because of such considerations, the integrated electronic health-care systems that evolve over the next two decades will surely contain elements of both the clinic-based and personalized approaches. Perhaps the health-care provider should offer patients the opportunity to take the patient-managed approach, but it’s not to be expected that all or even most patients will do so. Naturally, preferences vary not only personally but culturally and regionally. In Germany, for example, it proved impossible to introduce a standardized smart card with information concerning matters like emergency care, because of intractable disagreements about what specifically should be on the card. In developing its patient-centric HealthVault product, Microsoft has initiated research in Asia to find out how preferences about what belongs in a “comprehensive” health record differ from those in Europe and North America.
Ironically, those who would benefit most from the personalized approach—the elderly—are least well equipped with the computer skills to manage it. Though younger people have the savvy and can be expected to take responsibility for their records in time—but a long time!—it will probably take decades for patient-managed record keeping to become widespread. Meanwhile, however, mobile health apps—networked body sensors, emergency alarm systems, and so on—will become increasingly a part of our lives. Those systems will somehow be integrated into electronic health-care record keeping.
To a great extent, the fully computerized health-record-keeping system of the near future, however it is organized, will involve the wide application of existing technology. But the role of computers will not be confined merely to record keeping and the accurate transmittal of commands. As much more information is compiled and standardized, there will also be room for computers to perform analytic tasks autonomously, make suggestions, and offer physicians diagnostic alternatives.
The words used to describe conditions, procedures, and outcomes differ widely, however. So standardizing best practices and agreeing on common terminology is a nontrivial task. Even compiling accurate and complete tables of medical synonyms would not be a simple job.
Once much more information is compiled, integrated, and made universally available, however, computers can take on still more challenging tasks. They can comb the information for unsuspected connections, alert doctors to diagnostic possibilities, and warn them of hidden consequences. Some sense of this can already be seen in e-prescription systems, where computers are able to spot harmful drug interactions, detect histories of overprescription, and check the plausibility of dosage levels.
In addition to Microsoft’s efforts, Google, IBM (with its “information-based medicine” brand), and Oracle are exploring how informatics and cloud computing can be brought to bear on healthcare delivery problems. All the major IT companies are getting involved.
The drug discovery process will also benefit from computerized techniques and cloud computing. The process requires the storage and transmission of massive amounts of data. In the not very old days, pharmaceutical companies would literally deliver truckloads of data to regulatory agencies like the U.S. Food and Drug Administration. Now the agencies require all such data to be stored and submitted electronically. Cloud computing resources can be summoned not only to store data but to do searches of existing medical histories to determine which individuals might benefit from specific therapies or drugs. And applications can be found in the cloud to perform image analysis, a promising technique in molecule searching, which in turn is a big part of drug discovery.
Take the breast cancer known as HER2-positive, which can be stopped in its tracks if receptors on the exterior membranes of the cancer cells are somehow blocked by the right molecules. There are an immense number of candidate molecules to do the job, so why not look for them by means of massively parallel computing, available in the cloud? Such work, which used to take a year, may now be completed in hours.
is a professor in the departments of bioengineering and electrical engineering at the University of Washington, in Seattle. He is director of the university’s Image Computing Systems Laboratory. He has chaired SPIE medical imaging conferences and has served IEEE’s Engineering in Medicine and Biology Society (EMBS) in a variety of leadership roles.
is director of imaging infrastructure at Roche Pharmaceuticals, in Basel, Switzerland. She previously directed pharmaceutical imaging at Siemens. She has a doctorate in biomedical engineering and has been a leader in IEEE’s Engineering in Medicine and Biology Society (EMBS).