Hot and dry Dallas evening. After fighting the traffic and looking for a non-existing parking space, I finally made it to the event featuring a presentation by Dr. Andy Conrad, CEO of Google X Life Science. Eclectic atmosphere, sticky floors near the bar and dim lights fit perfectly to the occasion. I scanned the packed room and quickly noticed prominent scientists and biotechnology leaders mixed with graduate students, postdocs, and other attendees. The diverse audience made it clear that not only the scientific community was eager to hear the latest news from Verily (a new name for Google X), the most interesting and mysterious division under the Alphabet umbrella.
Finally the lights went down and Dr. Conrad’s presentation began. It was both strange and exciting to see the Google logo in life science publicity. Andy Conrad talked about several projects that have being developed by a newly assembled interdisciplinary team of talented biomedical (biologist, geneticists, clinicians) and non-biomedical (computer, social scientist, philosophers and even poker players) specialists. With a childlike amusement, I listened to the stories about contact lenses that continuously monitor glucose levels, a “smart” spoon for people with tremors, a disease-detecting nanoparticle platform, and a health-tracking wristband. These cutting edge innovations were complemented by the Baseline Study, an ambitious attempt to collect genetic and molecular human data in order to generate a model of a healthy person. The reality started to blur and I could almost taste the future.
The Baseline Study particularly drew my attention. We, scientists, strive to unravel disease mechanisms and dedicate years to figuring out what goes wrong inside an affected individual. Unfortunately, we can’t define with a 100% confidence what a healthy human body should be and how it should function. It is true that many clinical studies incorporate control groups that consist of healthy individuals. However, a large biological variability, different sampling, and analysis methods, often result in significant alterations in outcome measures between healthy participants from different research labs. Thus, the creation of a golden standard that could be unanimously used in clinical research is highly anticipated. No doubt, Verily is in a great position to undertake the arduous task to fill this knowledge gap. Why? Because Google is by far an industry leader when it comes to data collection, storage and analysis. The company has enough funds to sponsor a multi-centered longitudinal study and enroll thousands of people. More importantly, it has a long history of using machine learning to mine through terabytes of information.
What is “machine learning”, mentioned by Dr. Andy Conrad several times during his talk? Is it true that computers can learn? Yes, they can, and they do so by employing complex algorithms that can browse through data in search of patterns to generate predictions and/or make decisions. We come across the products of machine learning every time we glance at the Netflix or Amazon recommendations, or type an inquiry into a search engine. Interestingly, the same concept is now used in biomedical research. A growing field of evidence suggests that computers are capable of producing data-driven evidence-based diagnosis or medical decisions. For instance, Computational Pathologist Mode, or C-Path, is a machine based leaning method designed to predict patient survival by analyzing images of cancerous tissue. C-Path showed not only a correct cancer classification, but also demonstrated higher accuracy predicting survival than a human doctor. Another algorithms, a virtual interviewer, could identify human emotions by simply looking at pictures of human faces. The applications of this program are vast, spanning from developing new lie detectors to predicting psychotic or relapse episodes.
Despite the exciting possibilities, there are several concerns that should be addressed before we let machine learning do most of the thinking. The first concern is privacy. Even though Dr. Conrad assured the audience that the data will be anonymous with a “research only” tag, and compliant with all federal regulations, there were some raised eyebrows showing potential lack of trust in for-profit companies hosting such massive collections of complete personal and biological profiles. The second issue has broader societal implication. The deep machine learning allows non-medical specialist to perform various tasks that are currently done by medical professionals only. This trend, on one hand, will enhance the quality of healthcare, as doctors will have more time to spend with patients. On the other hand, it may threaten to eliminate many support staff positions.
At the end, it is hard to deny that health-related practices are metamorphosing into a novel technology heavy system. Not only Google, but several other tech giants such as Facebook, Apple and IBM are slowly entering the healthcare realm. So fasten your seat belts and get ready to witness a new era of healthcare and biomedical research based on the collaboration between computers and natural sciences, which promises the delivery of the unique personalized medicine of the future.
Editor: C. R. Morales