Jul 30 2015

Big Data and Personalized Medicine

Jun Wang, a famous Chinese geneticist, announced that he is going to shift his career into developing an AI (artificial intelligence) system that correlates genetics, behavior, and environmental factors with personal health. The goal is to provide individual recommendations about health and lifestyle based upon those factors.

In this case AI does not refer necessarily to a self-aware computer but just an intelligent system, like the AI that determines the behavior of characters in video games, or that won Jeopardy against human champions.

The real centerpiece of Wang’s vision is the data. He wants to build a database including the genomics data from one million people (and eventually much more), and correlate those genetic factors with lifestyle, environment, and health. What he is proposing, essentially, is using big data and AI systems to take the next step in personalized medicine.

Personalized medicine is currently a popular buzzword – you will find it frequently on alternative medicine sites. This is not because CAM practitioners are ahead of the curve. Rather, they latch onto the latest concepts and then make up the details as they go. It’s easy when you don’t have to do actual research or be science-based.

Real personalized medicine requires incredible data. The idea is that, rather than prescribing interventions based upon group data (the current standard) we individualize recommendations based upon data about the individual. Of course physicians already individualize treatment recommendations, as much as is possible. the idea, however, is to take this to the next level.

For example, there are many drugs used to treat hypertension (high blood pressure). There are certain health factors that guide the choice of which ones to use, but in the end there are likely several viable choices and practitioners use trial and error to find the best one for a specific patient.

What if, however, we could do a blood test to establish a genetic profile, and then based on the results choose the optimal hypertensive medication the first time. Such information could also be used to make recommendations about salt intake, optimal weight, the likely benefits of exercise, and other lifestyle choices. Right now we can make statistical statements about these things, but not necessarily individual statements.

This, of course, would be a huge advance in the practice of medicine, and health scientists know this and are working on it. But we are not there yet. The human body is fantastically complex, and the amount of data we would need to drill down to the individual level is massive.

That is where Wang comes in. Right now we are at the very beginning of personalized medicine. We are picking the low hanging fruit, where, for example, specific genetic factors might determine the optimal chemotherapy for a specific cancer.

Other tests are possible, like looking at liver metabolism to predict how a patient will handle certain drugs, but are not yet mainstream because the tests are not practical or cost effective.

The hurdle to adopting widespread individualized medicine is that we would need to look at more than genomics. What we need to know is not only what genes a person has, but how those genes have been translated into phenotype – the end result of biology. Genes manifest into phenotype based upon environmental factors, during development and even later. Genes interact also with lifestyle behaviors.

To really individualize treatments we would need to know not only a person’s genome, but their proteome, and perhaps their brain’s connectome, and other aspects of their biology.

Wang wants to create a database will all of this information for at least a million individuals, and then use AI systems to analyze the data to look for patterns. The ultimate goal is to collect similar information about an individual, so we can say that someone with these genetic alleles, this medical history, certain enzyme activity, and these behaviors would most benefit from a specific intervention.

This is ambitious. It is plausible, but this is the kind of project that will take a generation to really manifest and change the way we practice medicine. Meanwhile we will continue to make incremental advances in this direction.

If Wang is successful I do think his project has the potential to push the goal of  individualized medicine forward more quickly. He has the right idea – massive amounts of data intelligently analyzed is what we will need. He also seems to appreciate the complexity of the project. Mere brute force will not be enough. Biology is complex and medicine is hard, if you want to do it right.

2 responses so far