The explosive growth of wearable and surgically implanted sensors, combined with the ever increasing digitization of medical records has created massive growth in the data available to the healthcare and medical research industries. Wearable clinical devices are expected to surpass $6 billion in sales by 2019, and more and more people are sporting watches and smart phones that record heartbeat, and other vital signs throughout the day.
The result is an explosion of data points that reside in thousands of databases around the world. Within these data stores is a treasure trove of knowledge that may hold the answers to many of the most challenging questions facing modern medicine. However, all this data that resides in the traditional structured format of a database schema, is dwarfed by the unstructured data that exists within the tribal knowledge of healthcare professionals – the observations, diagnoses, prognoses, family histories, reports, published articles, and clinical studies that have been the traditional lifeblood of modern medical practice. Imagine the exponential increase in the value of the knowledge base we could create if we could blend this structured data with unstructured tribal knowledge on the fly, and query it using a simple natural language questions such as: What is the best long term diagnostic breast cancer screening regimen for women between ages 40 and 65 who are BRCA 1 and 2 negative? What are the optimal combination of drugs and their dosing regimens for treating type 2 diabetes in patients that are resistant to metformin?