Markets we serve > Life sciences > Intelligent data retrieval for COVID-19
Intelligent data retrieval for COVID-19
The COVID-19 crisis generated an extraordinary need to reduce the timeline from the identification of therapeutic drugs to the implementation of treatments that reduce the severity of COVID-19 and improve patient survival. Partnering with the University of California, Santa Barbara, we are addressing this challenge with Advanced Technologies for Acquiring Clinical Knowledge for COVID (ATACK) with the goal of developing, demonstrating and evaluating effective methods of answering medical questions by dynamically linking and integrating structured clinical trial databases with public research articles in multiple languages. We are exploiting the natural links between data derived from the same source, thereby establishing the validity of the data references, and developing software to answer users’ questions regarding the design and results of trials.
Allowing users to select predefined or pose ad hoc questions, the ATACK concept identifies answers in a collection of clinical trial data and articles. The key challenge lies in the question answering (QA) functions, which take questions as input and identify answers from the structured and unstructured source data.
We are exploring two complementary approaches for generating high-quality responses to biomedical questions that link structured and unstructured data. In the first approach, we use multi-label classification to optimize the solution to answer common, predefined biomedical questions. Our second approach uses a new deep semantic indexing method to support more open QA. Both of these approaches will make use of information extraction / retrieval technologies to extract structured data from databases and retrieve published research articles.
We are exploring two complementary approaches for generating high-quality responses to biomedical questions that link structured and unstructured data. In the first approach, we use multi-label classification to optimize the solution to answer common, predefined biomedical questions. Our second approach uses a new deep semantic indexing method to support more open QA. Both of these approaches will make use of information extraction / retrieval technologies to extract structured data from databases and retrieve published research articles.
The Principal Investigator and Manager of this program is Dr. Chumki Basu, a Senior Research Scientist at Perspecta Labs. Dr. William Wang, the Duncan and Suzanne Mellichamp Chair in Artificial Intelligence and Designs at UCSB, is a Principal Investigator. This work is sponsored by Office of the Director of National Intelligence/Intelligence Advanced Research Projects Activity (IARPA) as part of the COVID-19 Seedling Research Topics under the direction of Dr. Carl Rubino.
