Symptom-Based Disease Detection System In Bengali Using Convolution Neural Network

Natural language processing (NLP) and automatic detection of the disease have become popular in the recent era. Several research work show disease detection system in several languages. We present a disease detection system from the clinical text which is in Bengali language consisting of a numerous set of diacritic character, at a sentence-level classification. The clinical dataset consisting of Bengali text which is generally user interpreted symptom for the most common disease. Also, our approach represents the NLP methodology for Bengali language processing and classification of disease using several types of neural networks with hyper-parameter tuning and word vectorization. The aim of the research is the initial detection of disease from the user's voice to text data, in our case Bengali. So, a speech recognition system developed in the Bengali language is used to feed the disease detection model and finalizing the output with the model-detected disease.