NLP-based clinical text classification and sentiment analyses of complex medical transcripts using transformer model and machine learning classifiers

A foundation model in software design could be the fundamental framework or structure that a system is constructed around. This could comprise building blocks, libraries, or approaches to design that offer a strong foundation for creating more sophisticated software systems. A proposed software foundation framework illustrates the transformation of traditional models to potential advanced models. As a part of this objective, machine learning classifiers and pre-trained BERT and LSTM models are employed on a complex transcript dataset for context analysis and comparative of both models metrics is done to ML classifiers. The LSTM, a type of RNN, outperformed the ML classifiers in performing text classification from complex medical transcriptions. The accuracy achieved by LSTM is 0.94 with a precision of 0.87, and F1-score value is 0.90. SVM has an accuracy of 0.65 and a F1-score value of 0.64, whereas the convolutional neural network has an accuracy of 0.66 and F1-score value is 0.65. The metrics shows that LSTM, BERT models perform better to other ML classifiers and ensemblers in text classification of medical transcripts.