What is the future of personalized medicine
Two significant things happened this month that complement each other in highlighting the future of medicine. Earlier this month, the scientific journal Cell published a fascinating paper from Michael Snyder’s lab and others at Stanford University showing that information from “personal genomes” combined with information from individual electronic health records could accurately predict Abdominal Aortic Aneurysm (AAA), a severe cardiovascular disease, difficult to predict early, and highly fatal if left to develop to an advanced state.
In an earlier blog post, I noted that we are starting to see examples of improving treatment for every patient’s disease using information from the human genome sequencing effort. This recent publication also now shows us that even with for diseases with “unknown genetic underpinnings” like AAA – i.e., there’s genetics involved based on who develops the disease but the molecular evidence hasn’t been demonstrated yet – using advanced data analyses and available medical record histories can produce highly accurate disease predictors. Even more remarkably, the machine learning used to identify the genetic components and the associated clinical observations were “agnostic” to the type of disease. In theory, any disease with similar characteristics to AAA – i.e., pleiotropic with genetics a significant factor – can be studied to identify accurate genotype-phenotype predictors.
Also this month, Apple unveiled the fourth generation iWatch with new, health-related features. One feature – the ability to detect and record cardiovascular data as an FDA-approved electrocardiogram (ECG) device – represents the type of longitudinal data capture that will create opportunities for incredibly rich, data-driven analyses of human heart health over time. With Apple’s ResearchKit and the growing availability of genome sequences for individuals, we may be able to collect massive amounts of data that will inform studies like the AAA work, and lead to the identification of subtle indicators of disease states previously unknown.
We still lack the robust information systems needed for the genomic medicine to become commonplace. Where will the data be kept and how will we access it? What data analyses can be done on an individual level, and when will we know the analyses are appropriate and accurate? As with the history of several transformative technologies, the systems we have today will likely need to catch up with innovation.
September 27, 2018