Health-related information is the new frontier of data, and machine learning is one of the most powerful data-analysis tools available. When you put them together, you can glean valuable insights about health and human disease. JDRF and IBM announced today they are collaborating to focus this power on preventing the autoimmune disease type 1 diabetes (T1D).
In T1D, a person’s pancreas stops producing insulin, a hormone people need to get energy from food. Approximately 1.25 million Americans have T1D today, and the incidence is growing. T1D can strike people suddenly, at any age, no matter their diet or lifestyle. T1D’s causes are not yet entirely understood, but we believe there are both genetic factors and environmental triggers involved. There is currently nothing anyone can do to prevent it, and there is no cure.
JDRF is working to change all that. We are the leading global organization funding T1D research, inspired by our vision of a world without T1D. We are excited to bring IBM’s world-class computing power to bear on untangling the complex interplay of factors involved in the development of this terrible disease.
Research so far has shown that T1D develops differently in different people. For example, we know factors like a person’s age can influence disease course. We also know that T1D progresses through a series of defined stages, and we have supported multiple, long-term studies tracking disease progression in different groups of people around the world. This has yielded detailed timelines of disease course in tens of thousands of people, along with records on family history of T1D, genetics, other medical history, environmental factors, and diet.
The potential of this information is remarkable! It holds the key to important insights into T1D development that could eventually enable us to prevent or at least delay its onset. But to unlock its full potential, we need to view the data holistically. Unfortunately, the data sets are independent, having been collected in different ways, at different times, in different locations, by different people.
Researchers have begun to pool the data sets for analysis. In 2013, a team analyzed pooled data from children genetically at risk for T1D collected over one or two decades during three studies in the US, Finland and Germany to determine the rate of progression of T1D from the earliest stage to full-blown disease . They found that most of the children progressed to full-blown T1D within 15 years and that the rate of progression was affected by age, genetics and sex.
We know there is more to learn—and we need more computing power to do it. That’s where IBM comes in. Collaborating with IBM gives us the power to unite and analyze multiple data sets for insights on T1D development that will lead to the development of strategies to stop it in its tracks.
JDRF and IBM scientists will analyze at least three previously collected data sets from global research projects and apply machine-learning algorithms to find patterns and factors at play. First, they will compare, clean up and combine all the data. They will then look for patterns in the combined data and identify various subsets of T1D. Finally, the team will develop models describing progression to full-blown T1D in the various subsets. This large-scale data analysis will lead to deeper understanding of the risk factors and causes of T1D and, eventually, finding a way to prevent T1D entirely. The knowledge we gain could also help lead to a cure for those who already have T1D.
IBM’s technical capabilities and computing power make a great match with JDRF’s connections to research teams around the world and subject matter expertise in T1D research. By teaming up, we can piece together the puzzle of T1D.
Learn more about JDRF’s work in diabetes research by visiting their website
Check out the latest about IBM Research’s Healthcare & Life Sciences work
 JAMA. 2013;309(23):2473-2479. doi:10.1001/jama.2013.6285