Sometimes a single patient inspires a question. For Fiona, the severe burn of an 8 year old boy who died a few years later of cancer started her on a journey of discovery. She asked if burn injuries are linked to downstream occurrence of cancer and other diseases. After exploring this question using linked EMR data sets in Australia, our research team found a positive association between patients who had suffered burn injuries and higher rates of cancer, cardiovascular disease and other conditions.
Then we asked what hidden biological links existed between burn injury and cancer. What genetic changes take place during a burn injury? Is there a link between burn scars and tumor fibrosis? Does inflammation and infection play a role? With IBM Watson for Drug Discovery, Fiona and our team are beginning to explore these linkages to understand why burn patients are at higher risk of developing other diseases such as cancer.
IBM formed Watson HealthTM in 2015 with the aspiration to improve lives and give hope by delivering innovation to address the world’s most pressing health challenges through data and cognitive insights. WatsonTM for Drug Discovery (“WDD”) is a cognitive discovery platform that helps accelerate breakthroughs in bench and translational science. Watson reads over 29 million MEDLINE abstracts, all the Open Source PMC full text journal articles, over one million licensed journal articles and U.S. and European patents to help speed disease understanding, target identification and drug repurposing. Using predictive analytics, WDD also ranks and propose novel gene, disease and drug candidates to help researchers prioritize promising leads for further research.
Now our team is using WDD to help researchers understand the biological mechanisms that occur during burn injuries which may be linked to cancer and other diseases such as cardiovascular, respiratory and gastrointestinal conditions - allowing the team to interrogate the literature in ways we never could previously.
Take for example the network graph below. To answer the question, what genes have some relationship to burns, fibrosis and cancer, a researcher could simply type, “Burns,” “Fibrosis” and “Cancer” into Watson for Drug Discovery and within a few seconds Watson surfaces over 2 dozen genes that are mentioned in relation to each. Watson can also extract the relationships between genes such as whether they upregulate or downregulate each other and can filter genes by a biological process they are related to like inflammation. Identifying genes that overlap between these areas using traditional search engines would likely take a researcher much longer. It might require researchers to search for synonym names for each gene and search for multiple types of relationships and functions. Watson can read and extract these relationships at scale as well as present them all simply, in a way that provokes novel hypothesis about possible linkages.
WDD extracts all the genetics relationships mentioned to diseases from this text and present them in a simple way that provokes novel hypotheses about the origins of disease and connections between conditions that one may have never considered.
Through the lens of burn injury, Watson may help our research team discover some of the mechanisms of cancer.
Our hope is that Watson will facilitate the identification of novel approaches in burn care that will help reduce the risks of developing other downstream illnesses.