How will medicine and its regulation adapt to the information age?
This commentary in the Genomics Law Report’s ongoing series What ELSI is New? is contributed by Andro Hsu, 23andMe, Inc.
Jorge Luis Borges once compared scientific endeavor to the making of a map as large as the territory the map depicted. As the price of generating DNA sequence and other data types decreases more quickly than Moore’s Law, we are well on our way to creating a map even larger than reality.
To handle these vast amounts of biological data, we will eventually need computers not only for storage, but also to run algorithms that can simultaneously process, synthesize, and evaluate thousands of interacting biological and environmental variables―tasks beyond the capabilities of any human brain.
The problem? Our notion of evidence―and the regulatory framework that governs its use in setting clinical guidelines, approving drugs and medical devices, and making medical decisions―was developed during a reductionist era of low-hanging informational fruit, when tiny sample sizes sufficed to demonstrate large relative risks. Today’s holistic approach uses higher-resolution, system-wide data to power sophisticated statistical models that produce incremental gains in prognosis, but require ever larger samples to validate.
So what happens when we can obtain so much unique information about a person that the only appropriate population for validating predictions made from those data is the subject herself? Conversely: even if we had comprehensive biological and environmental data for the seven billion humans in the world, how could we provide sufficient evidence when rigorous validation of all possible interactions requires a sample size larger than the world itself?
To benefit from increasingly detailed data, we need to cooperate in finding a way for regulators, health care professionals, and patients to get comfortable with multivariate algorithms, computational models incorporating correlations without underlying causation, and tests that measure and combine analyte levels to produce individualized answers in the form of probabilities. Welcome to systems biology, a field as messy as life.
I have an answer to my own question! Lee Hood gave an excellent talk at the Burrill & Co. personalized medicine meeting. He speculated that in the future we will all get a twice-yearly scan of organ-specific blood-based protein biomarkers that are identified as surrogate endpoints for disease.
If this happens, each person will be his/her own control!