08 Apr Q&A with Eric Topol, Executive Vice President, Scripps Research Institute
Dr. Eric Topol, voted by Modern Healthcare as the most influential physician executive in the United States, shared his thoughts on why people should own and control their own data.
Dr. Topol, a world-renowned cardiologist, geneticist, digital medicine researcher, and prolific author, is a true pioneer in personalized medicine. In 1975, he wrote his baccalaureate thesis, “Prospects for Genetic Therapy in Man” earning him an undergraduate degree with highest distinction from the University of Virginia. His work in the genomics of heart attack has led to the discovery of key genes and resulted in recognition by the American Heart Association as one of the top 10 research advances of the year, twice. Topol is Founder and Director of the Scripps Translational Science Institute. In 2016, Topol was awarded a $207 million NIH grant to direct a large portion of All of Us (Precision Medicine Initiative), the one million American prospective research program). Read his full bio.
Dr. Topol will present and be honored at PMWC 2018 Michigan taking place June 6-7, 2018.
Q: Genomics, Digital Health, Big Data, and Artificial Intelligence (AI) are some of the newest technologies/fields that are reshaping medicine and healthcare. How and when will they impact healthcare?
A: Right now, we have extraordinary potential to understand the medical essence of human beings in an unprecedented fashion. Surrounding that, there is considerable hype, while we also see some real advances – so there is a mixture of substance and hyperbole. Yet, it is undeniable: sensors, genomics, imaging, advanced analytics and particularly deep learning, a major subtype of artificial intelligence, are all making big strides. They are clearly going to make an impact on the future of medicine.
The progress is quite heterogeneous, with respect to medical conditions, timing ,and who will be affected. If you look at the person, the consumer, these technologies will be able to help manage a condition like diabetes. But today, the algorithms are dumb and only inform a patient that their glucose is going up or down, and they don’t factor in anything else in their lives. We now have far better glucose sensors that are factory calibrated, that will eventually, at lower costs, will make finger-sticks obsolete with smart algorithms. These glucose data will be integrated with other data: your activity, your sleep, your levels of stress, your nutrition, your gut microbiome, and much more. That will help coach people to regulate their glycemic levels far better than what we do today, and we will transcend imprecise parameters like glycohemoglobin. This is just one example of a new data and analytic framework for diabetic patients, or even people simply at risk for diabetes. The same kind of example would apply for high blood pressure or any other common chronic conditions. Eventually, but this will take much longer, we will see this across a person’s health –the approaches will be holistic. This is all on the individual level, but there are obviously many other levels (for doctors, hospitals, etc) where big data, AI, and the digital genomic revolution are going to have a significant effect.
Q: Is it feasible to successfully implement individualized/precision medicine for everyone and what are the biggest challenges?
A: Yes, it is feasible. Our number one problem today is the inability for people to have all of their data. No one has all of their medical data from birth, or even prenatal, all the way through life. Every medical encounter, every lab, every scan – no one has all that data, unlike places like Estonia and other countries where patients have and own all that data. This is step number one and currently we are still far away from that, but eventually that has to occur to maximize the data’s impact and to help preserve one’s health. If we are going to work with very limited data it is going to be a serious impediment in the area of AI. Instead of health systems and doctors owning the data, people need to own their data– it is vital that this flip occurs. And patients will very happily share their data with specific health providers, doctors, and others. But they need to be the ones who are making that decision.
Q: Some of these approaches to personalized medicine require individuals to continuously monitor a range of personal readings, such as blood pressure, glucose levels and heart rate. In light of recent renewed concerns regarding data security and patient data privacy, how will people’s minds and the general culture be changed to trust the collection and sharing of personal data with providers, via wearables or smartphones?
A: This is the number one issue for me, which will hold us back–privacy and security of the data. The endless sea of data breaches – whether it is Facebook or MyFitnessPal – can’t be tolerated in the medical world. This is another reason why people have to own their data and on a secure platform, like Blockchain or a private cloud. We need to achieve the highest level of data security and personal ownership. If you ask any cyber security guru for the best strategy, in relation to avoidance of data breaches, the first step is getting the data out of massive servers and getting it into units of one or small family units. Massive data servers are incredibly attractive to cyber thieves and hackers and medical data sells for 5-fold more than personal financial data on the Dark Web. This is yet another reason why people must own their data and have it made fully secure.
Q: How can we accelerate the developments, and deliver on the promises of precision/individualized medicine?
A: Give people control over all their data and let them own it. But we are a long way from that. We are seeing some first instances with Apple enabling people to get their data from health systems that have a portal and are cooperating with Apple. This is a good start. And that happened quickly, so for example, I got some of my data from both Scripps Health and UCSD Health, so for the first time I have that and my sensor data in one place, at my fingertips. There’s plenty of missing data, like scans, and multiple providers, but it’s far better than trying to go to multiple, clunky health system portals. This is the beginning of a quiet revolution for people – not just having their data, but eventually also owning it. This is a new development that will continue to evolve and be built upon, and this is important. It’s a signal of what is to come. We can’t do deep learning AI for any individual optimally without all of their data.
Besides data security the biggest problem are the deep learning algorithms that are so remarkable and can read scans, read slides, transform speech to text, etc. They all have to be validated, prospectively, in the clinical environment. What we have today is lots of in silico analysis and retrospective validation, but that is not good enough to change medicine. We need to see all of these things, whether it is retinal diagnosis, an EKG, or a heart arrhythmia, any type of medical scan, cancer pathology slides, or skin lesions—validated prospectively so that the medical community can accept it. None of this has been done to date. There isn’t one prospective clinical validation of these promising deep learning algorithms yet. At least nothing that has been published as of April 2018.