Predição de doenças usando transformadores para conjunto de dados de expressão gênica

Introduction: Gene expression is a significant process which bridges the gap between information encoded within a gene and the final functional product of a gene. Therefore, evaluating gene expression is a vital process in terms of developing treatment and monitoring of disease, as diseases can result in unforeseen consequences. Objectives: Various studies have incorporated the Deep Learning (DL) concept with gene datasets for prediction of disease. However, prevailing DL methods are considered to be extremely ineffective in terms of accuracy and other aspects of disease prediction. Therefore, DL with transformers are applied for predicting the disease with gene dataset. As transformers are self-attention mechanisms, this permits the use of contextual information for any location in the input sequence and aids in capturing long range dependencies, which aids in delivering better accuracy and effective model for predicting disease. Methods: This paper focuses on the applications of gene expression with different DL algorithms along with transformers for prediction of diseases. Some of the applications of transformers with gene expression dataset include prediction of cancer such as lung cancer, stomach cancer and drug discovery, a focus of this paper. Results: Crucial analysis is undertaken by considering certain aspects such as architecture, dataset, process. This analysis helps in finding the aspects of the DL algorithms with and without transformers. Finally, gaps are identified through the analysis of transitional researchers and could be considered as future recommendations by overcoming the gaps that are intended to create promising work in this area. Conclusion: Despite having several advantages, there have been limited studies in terms of prediction of disease. Different applications such as cancer prediction, drug discovery are looked at in the present study. It has been observed that incorporation of transformers in DL algorithms were implemented more beginning from the year 2021; still there are a few complications which needs to be overcome for prediction of diseases. DL using transformers alongside gene datasets will be helpful in disease prediction. This study would aid researchers in their innovations in this area for enhancing the effectiveness and efficiency in prediction of disease.