Abstract
Over the decades, diabetes has proven to be a chronic disease,
causing significant impact on individuals and healthcare systems
globally. This disease increases the threat of diseases like cardiovascular
illness, blindness, and may even cause early death. Diabetes
requires a lifelong disease control, taking a significant financial
toll on, not just the patient, but the entire family. Of the global
diabetes cases, 90% are Type 2 diabetes mellitus (T2DM). By taking
preventive measures, like early diagnosis and with proper care, the
effect of diabetes can be decreased, possibly delaying its complications.
This paper studies literature on implementation of machine
learning (ML) and deep learning (DL) approaches for predicting
diabetes and evaluate their performance. The accuracy of diabetes
prediction using Support Vector Machine (SVM) algorithm and the
Artificial Neural Network (ANN) algorithm, on the Pima Indians
Diabetes (PID) dataset, was compared. ANN showed better accuracy
as compared to SVM. Also, the Adam optimizer proved to be a
better predictor than RMSprop optimizer. The results suggest that
machine learning and deep learning can be an effective tool for
predicting diabetes, and that some algorithms perform better than
others. We conclude that these techniques have shown accurate
results in prediction of diabetes. Future studies should extend this
model using more neurons in the hidden layers of ANN with a
focus on developing a robust model that can be easily integrated
with clinical practice.
causing significant impact on individuals and healthcare systems
globally. This disease increases the threat of diseases like cardiovascular
illness, blindness, and may even cause early death. Diabetes
requires a lifelong disease control, taking a significant financial
toll on, not just the patient, but the entire family. Of the global
diabetes cases, 90% are Type 2 diabetes mellitus (T2DM). By taking
preventive measures, like early diagnosis and with proper care, the
effect of diabetes can be decreased, possibly delaying its complications.
This paper studies literature on implementation of machine
learning (ML) and deep learning (DL) approaches for predicting
diabetes and evaluate their performance. The accuracy of diabetes
prediction using Support Vector Machine (SVM) algorithm and the
Artificial Neural Network (ANN) algorithm, on the Pima Indians
Diabetes (PID) dataset, was compared. ANN showed better accuracy
as compared to SVM. Also, the Adam optimizer proved to be a
better predictor than RMSprop optimizer. The results suggest that
machine learning and deep learning can be an effective tool for
predicting diabetes, and that some algorithms perform better than
others. We conclude that these techniques have shown accurate
results in prediction of diabetes. Future studies should extend this
model using more neurons in the hidden layers of ANN with a
focus on developing a robust model that can be easily integrated
with clinical practice.
| Original language | English |
|---|---|
| Title of host publication | Proceedings of the 5th International Conference on Information Management & Machine Intelligence |
| Publisher | ACM Digital Library |
| Publication status | Published - 13 May 2024 |