Open Access

ARTICLE

Deep Learning Model to Detect Diabetes Mellitus Based on DNA Sequence

Noha E. El-Attar^1,*, Bossy M. Moustafa², Wael A. Awad³

1 Information Systems Department, Faculty of Computers and Artificial Intelligence, Benha University, Benha, 13518, Egypt
2 Mathematics and Computer Science Department, Faculty of Science, Port Said University, 42525, Egypt
3 Computer Science Department, Faculty of Computers and Information, Damietta University, 34517, Egypt

* Corresponding Author: Noha E. El-Attar. Email:

Intelligent Automation & Soft Computing 2022, 31(1), 325-338. https://doi.org/10.32604/iasc.2022.019970

Received 03 May 2021; Accepted 08 June 2021; Issue published 03 September 2021

Abstract

DNA sequence classification is considered a significant challenge for biological researchers to scientifically analyze the enormous volumes of biological data and discover different biological features. In genomic research, classifying DNA sequences may help learn and discover the new functions of a protein. Insulin is an example of a protein that the human body produces to regulate glucose levels. Any mutations in the insulin gene sequence would result in diabetes mellitus. Diabetes is one of the widely spread chronic diseases, leading to severe effects in the longer term if diagnosis and treatment are not appropriately taken. In this research, the authors propose a hybrid deep learning model based on Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) to classify the DNA sequence for the insulin gene to predict type 2-diabetes based on gene sequence mutation. To evaluate the proposed models, we used several performance indexes such as accuracy, precision, sensitivity, recall, and F1 score. The experiments shown in the paper reveal that the proposed model accomplished the best results. The overall accuracy of the learning process is recorded as 99% for the proposed hybrid LSTM-CNN model while it is recorded as 97.5%, and 95% for CNN, and LSTM, respectively.

---

Keywords

---