Performance Evaluation of Clinical Models on Sequential Clinical Text for AI Powered Medical Applications

Clinical terminologies are used by the experts to evaluate or conclude specific results based on the symptoms of the patients. The clinical terminologies used by the experts can be termed as aspects, aspect extraction is the major sub-task in any sentimental analysis. For the extraction of aspect from a given sentence, the existing natural language processing methods mainly focuses on the results generated by a pre-trained model. These pre-trained models work very well for normal English, however when it comes to clinical terminologies there exists a deficiency in generating the specific results. To address the issues related to clinical terminology. There exist multiple clinical models such as Clinical Big bird, Clinician Bert, Clinical Longformer which are transformer-based learning model which perform better against these clinical data. These models perform better on large dataset. In this paper, we would like to evaluate the performance of these Bio models with sequential clinical text and compare the performance of these models and evaluate context and coherence. The best results of these models will be presented to artificial bee colony for further optimization and integrate with an Artificial Intelligence based applications.