Dr. Eric Topol: Toward a personalized, learning healthcare system

The great inversion of medicine, with its roots just starting to take hold now, will have been fully achieved over the next few decades.

Instead of pervasive medical paternalism that dates back centuries, each individual will have much more than access to their medical data—they will be generating and owning it.

The smartphone, which engendered a revolutionary change in how we go about our lives, will ultimately replace (with some key hardware attachments) the stethoscope as the iconic symbol of medicine. It will be used to capture biosensor real-time, real-world physiologic data from any specific organ or system of the body, to run routine labs and sequence potential pathogens, to perform medical-grade scans of one's body, to track environmental exposures and food ingredients, to do most of the physical exam, and to connect with a doctor or healthcare professional at any moment in time.

But this pocket device capability will be much bigger than just amassing multidimensional data. No human being could assimilate the continuous flow of multiscale, torrential data, which, in contrast, is perfect for machines to process using artificial intelligence and deep learning from and for each individual. The output will be like that of supercomputer IBM Watson, only to the 4th power, with algorithmic recommendations yielding the exciting potential to pre-empt many acute medical conditions—like asthma attacks, heart attacks, seizures, arrhythmias and heart failure.

Such external wisdom of body guidance will be constantly fed back to the person using their choice of voice, text or avatar communication. Just as most people have so quickly and heavily come to rely on their smartphones today for the purveyor of all-important information, health and medicine will have been fully integrated.

This power of information and machine support will have a striking effect on the patient-doctor relationship, along with how clinics and hospitals function. Since the patient is autonomously generating most of the data with validated algorithmic interpretation, the doctor's role will shift to providing oversight, mapping out therapeutic and preventive strategies, and tapping into one's knowledge base, experience and wisdom. Embracing the shift of much responsibility for data collection to patients, and its interpretation by artificial intelligence, the human factor of the doctor—establishing trust and support with extraordinary communicative skills and real intelligence—will be indispensable.

The physical office visit will be unusual, save for very important interactions such as the discussion of a new, serious diagnosis. Instead, comprehensive data exchange, far greater than today, will take place virtually along with face-to-face discussion. Both the patient and doctor will be equipped with augmented reality to see all of the data visualization and predictive analytics. The record of the visit will be created via natural language processing, edited by the patient and the doctor (the latter with the help of machine learning to accelerate it). The idea of using keyboards or human scribes will be as foreign as pre-Gutenberg printing using scribes.

Hospitals will undoubtedly be used quite differently than today. While intensive-care units, operating rooms, emergency care centers and sophisticated image equipment will be maintained, the rest of the hospital functions will be exported to the home. Data surveillance centers will be the norm to remotely monitor large numbers of people in a community, staffed by healthcare professionals to intervene when necessary. Of note, the hospital information system of today will be considered a relic.

After all the billions of dollars that have been put into health information systems in hospitals, how could these no longer be used or required? The answer lies in the need to completely decentralize the data to units of one or a few, rather than aggregate people's data into thousands or millions. Coincident with this decentralization is the information power shift that sets up individual ownership of all of one's data—the bulk of which will be generated by that person.

A peer-to-peer network, such as via a blockchain model, will be used to provide each person their encrypted data—in an electronic wallet or locker—to share when and with whom the individual chooses. Finally, the civil right of individuals to own their medical data will have been actualized through governmental legislation.

One more thing that we'll see by then: the planetary medical knowledge resource. Medical data-sharing to help one's fellow man will be the norm when individuals have outright ownership of their data, without concern over a breach of privacy or re-identification. All the big data per individual can be collated for more than a billion people (someday billions) and nearest-neighbor matching analysis from this resource will be critical for selecting the best prevention and treatment options. We will have reached the era of datafied, individualized medicine and a true, self-perpetuating learning health system.