A Study of Feature Extraction Methods and Corpora in Developing a Deep Neural Network Model for Contradictory Claims Detection in Medical Literature

Clinicians or clinical researchers may encounter doubts or questions related to clinical tasks or decision-making. To seek the answers to their questions, they will search and evaluate the best current evidence from medical literature through electronic resources such as online databases. However, research claims that answer the same clinical question often contradict each other. The existence of contradictory research claims, in addition to the huge number of literatures to be appraised, can make evidence appraisal process challenging. This research explores various techniques in building an artificial intelligence model to detect contradictory research claims to support the clinical evidence appraisal process.