Improving Eligibility Classification on Clinical Trials Document using Bidirectional Long Short Term Memory Recurrent Neural Network

Cancer clinical trials intervention are generally too restrictive, and some patients are often excluded on the basis ofcomorbidity, past or concomitant treatments, or the fact that they are over a certain age. In this research we built aclassification model for clinical information using public clinical trial protocols labeled as eligible or not eligible. Textclassifications are trained using deep learning to determine the predictive outcome of short free text statementsreflectingeligible and not eligible clinical information.then we also performed semantic analysis for the obtained wordembeddingrepresentations and were able to identify similar treatments. We have proven that learning outcomes using deeplearning methods to extract medical information from clinical trial documents have been successful in assisting healthpractitioners in prescribing treatments. The evaluation results showed a value with an accuracy value is 77.74%, precision is76.8%, recall is 80.80%, and F1-score is 78.80%