IBM, JDRF partnership using machine learning methods to tackle Type 1 diabetes

By Jeff Lagasse
Share

IBM and JDRF, a global organization funding Type 1 diabetes research, are joining forces. The collaboration will develop machine learning methods and use them to analyze years of Type 1 research data to pinpoint the factors that can contribute to the onset of the condition in children.
 
Type 1 diabetes  -- T1D for short -- affects about 1.25 million Americans, and to date does not have a cure. What the research collaboration will attempt to do is create an entry point in the field of precision medicine -- combining JDRF’s connections to research teams around the globe, and its subject matter expertise in T1D research, with the technical capability and computing power of IBM.
 
“At JDRF, we are absolutely committed to seeing a world without type 1 diabetes, and with this partnership, we’re adding some of the most advanced computing power in the world to our mission,” said Derek Rapp, JDRF President and CEO, in a statement. “JDRF supports researchers all over the world, but never before have we been able to analyze their data comprehensively, in a way that can tell us why some children who are at risk get T1D and others do not. IBM’s analysis of the existing data could open the door to understanding the risk factors of T1D in a whole new way, and to one day finding a way to prevent T1D altogether.”
 
IBM scientists will look across at least three different data sets and apply machine learning algorithms to help find patterns and factors that may be at play, with the goal of identifying ways that could delay or prevent T1D in children.
 
In order to match variables and data formats and compare the differing data sets, the scientists plan to use previously collected data from global research projects. Data analysis will explore the inclusion of genetic, familial, autoantibody and other variables to create a foundational set of features that is common to all data sets. The models that will be produced will quantify the risk for T1D from the combined dataset using this foundational set of features.
 
As a result, JDRF will be in a better position to identify the top predictive risk factors for T1D, cluster patients based on top risk factors, and explore a number of data-driven models for predicting onset.
 
“Nearly 40,000 new cases of type 1 diabetes will be diagnosed in the US this year,” said Jianying Hu, senior manager and program director at the Center for Computational Health at IBM Research, in a statement. “And each new patient creates new records and new data points that, if leveraged, could provide additional understanding of the disease. The deep expertise our team has in artificial intelligence applied to healthcare data makes us uniquely positioned to help JDRF unlock the insights hidden in this massive data set and advance the field of precision medicine towards the prevention and management of diabetes.”
 
The collaboration will unfurl in phases: In the near future the goal will be furthering the analysis of big data to better understand the causes of T1D. After that, the collaboration may also start to analyze more complex data sets, such as microbiome and genomics or transcriptomics data. Ultimately, JDRF hopes all of this will someday result in a cure.