Health and medicine will undergo a greater transformation than any other industry or field in the next decade.
At this year’s Exponential Medicine 2014, the overriding theme for the event was information. In his opening talk, Peter Diamandis said health and medicine are poised to undergo a greater transformation than any other industry or field in the next decade. Of course, he meant treatments and technology will meaningfully advance. But more than that, it is the liberation of data that will make care more targeted, proactive, and effective.
To understand the future, however, it’s critical to understand where we are now.
Venture capitalist Vinod Khosla wrote way back in 2012 that modern healthcare is more about the “practice of medicine than the science of medicine.” Diagnosis and treatment are more art than most will admit, and this is problematic because, by definition, 50% of all doctors are below average practitioners—acceptable in art, frightening in medicine.
Tens of thousands of ICU patients die annually due to misdiagnosis. Go to three different doctors and you’ll get three different diagnoses and plans for treatment, Khosla wrote. This isn’t to slander doctors, but to say most are faced with an impossible task.
Further, just as today’s doctors make life or death decisions on extremely limited information, researchers and scientists similarly draw broad conclusions from small datasets, a tiny slice of the population over a short period of time. Indeed, in her keynote on clinical trials, Dr. Laura Esserman noted that 70% of clinical trials fail.
This is likely, in part, because the studies informing those trials are not backed by information over broad populations but are instead handcuffed by over-specificity and a dearth of data.
Today, information isn’t free. But liberating forces are massing on the horizon.
Sensor technology, of course, is front and center. A profusion of body sensors are poised to be strung throughout the environment and in and on our bodies. These sensors are tiny, cheap, energy efficient, and most importantly, connected.
Sensors stand to collect information, not once every year or two, but every day, hour, or minute. They can open a window on disease before it becomes critical, before symptoms drive us to seek help, making diagnoses early and more accurate.
The best known health devices adapt smartphone motion sensors to detect movement (e.g., step trackers). But these are just the beginning. The next wave of sensors will measure a range of vital signs connected to the heart, blood, and brain.
Sensors on display at Exponential Medicine included two elegant EEG devices for recording brain activity, the Muse headband and iBrain. And the winner of the XPRIZE Nokia Sensing Challenge, awarded at the conference, is a compact system capable of running a wide range of diagnostic lab tests with a single drop of blood.
Just as sensors begin collecting new information, we may begin unlocking and leveraging already existing data within the system. Hospitals alone offer a wealth of information which is invisible to patients and doctors alike.
The system has all but scrambled this information, but data scientists are showing how software can piece it back together and make it useful.
Dr. Isaac Kohane told the story of a group of patients seeking recurrent treatments for various injuries. Using software to analyze the pattern of treatments, Kohane made a surprising diagnosis—domestic abuse. Indeed, it was later reported that these patients were victims of abuse, but not until well after they’d been released from the hospital.
Kohane believes a lot more such information exists within hospitals, if only anyone cared to look.
In addition to doctors, researchers may use information from sensors and the system itself to study populations of tens or hundreds of thousands of patients. And these studies will cover periods of time before, during, and after disease strikes.
The famous Framingham heart study collected information every few years from a few thousand patients over several decades. Framingham yielded profound insights into cardiac disease. Now, imagine doing the same study again—only collecting information every day and expanding the study’s population by an order of magnitude or more.
The Health eHeart study, spearheaded by UCSF’s Dr. Jeff Olgin aims to do just that. Health eHeart shows not just what’s possible in the future study of heart disease, but in the study of all disease. Broad, detailed data may soon be the rule.
Making It Meaningful
Doctors are already overwhelmed by the flow. Keeping up with a body of research that doubles every five years is a herculean task—perhaps an impossible one for mere mortals. How will we fare when information exponentially increases?
As genomics and synthetic biology pioneer Craig Venter said in his keynote talk, data isn’t the goal. The bigger objective is taking the data and making knowledge of it. How will we do that? Artificial intelligence.
Vinod Khosla believes computers will replace up to 80% of the tasks doctors perform today. This will result in significantly fewer errors, lower cost, less work per doctor, faster interactions, and more opportunities for doctors to do research.
But, as Exponential Medicine executive director Daniel Kraft noted: We shouldn’t think of it as AI but IA—intelligence augmentation. In the future, doctors will pair up with intelligent software to more quickly and comprehensively research, diagnose, and prescribe treatment plans.
IBM’s Watson, for example, is able to scan a field’s entire body of up-to-date medical research in fractions of a second and turn up relevant studies, rare drug side effects, even potential diagnoses. And as Watson searches text, machine learning techniques are equipping software with the ability to scan images.
Jeremy Howard, Founder and CEO of Enlitic and previous Chief Scientist at Kaggle, said the accuracy of object classification—identifying discrete features in images—has undergone massive improvement in the last several years. Already these algorithms are proving themselves superior to humans in the analysis of some cancerous tissues.
The convergence of these techniques will help us better manage all this new information—whether it’s finding causative correlations in genomic research or making more accurate, timely diagnoses in the doctor’s office and hospital.
But perhaps the most powerful effect of intelligent software on medicine? As machines do what they do best, doctors can refocus on what humans do best. Less overwhelmed by data they can’t possibly digest, doctors will find time to build relationships with the patient—answering questions, keeping them informed, making them comfortable.
The Dark Side
As more patient data is collected and made available for study and diagnosis, and more devices connect to the internet, health information will present a target for exploitation—if it’s online, it’s hackable.
According to Marc Goodman, typical identity theft is worth $2,000 to the thief—medical identity theft is worth more like $20,000. So far in 2014, medical cybercrime is up 600% because, Goodman says, it’s an easy target.
The answer isn’t to halt innovation but to pay more attention to security and enforcement. Goodman suggests some simple solutions: switching passwords on every website, securing connections to public networks, data encryption—and most critically, perhaps, taking care what information is shared online.
Technology as a Tool
Technology is amoral, it’s what humans do with it that determines whether it is a force for good or evil. In the coming years, we’ll have ample opportunity to adapt to a world awash in health information. We may decide to place severe limits on what and how information is shared. But the likelier outcome? The benefits of information sharing will outweigh the risks.
“We’ve gone from a data-poor world, to a data-rich world,” Larry Smarr told participants. “I’ve been through a lot of fields in my life. This is about as excited as I’ve been for research and what it’s going to do to change our lives.”
Via Singularity Hub