Language model-based deep learning for automated disease prediction from symptoms

Abstract

Effective medical care depends on accurate and prompt disease prognosis based on patient symptoms. In this study, we suggest creating deep linguistics models that given a user's brief description of symptoms, can correctly predict the condition. We want to increase the effectiveness of disease prediction by utilizing deep learning technologies, leading to better patient outcomes. We demonstrate the efficacy and validity of our suggested techniques of disease prediction based on symptom reports through extensive trials and evaluation. A full grasp of the disease state and its corresponding symptoms is necessary in order to accurately predict disease from symptoms. Traditional methods frequently rely on manual diagnosis by medical experts which can be laborious and arbitrary. With the help of recent developments in deep learning, it is now possible to automate and enhance disease prediction based on symptom descriptions. In this paper, we suggest creating deep learning language models that, given a user's brief description of their symptoms can reliably predict illnesses. By learning from a lot of medical data, our model tries to identify complicated correlations between symptoms and diseases. By adopting deep learning architectures, we can leverage the ability of models to extract meaningful representations from text data and make accurate predictions. In our research paper, DistilBERT achieves the highest accuracy surpassing Ensemble classification with an accuracy rate of 93.3%.