| Autor: |
Junaid, Sahalu Balarabe, Imam, Abdullahi Abubakar, Mohammed, Abdullahi, Teck Ching, Daphne Lai, Haji Naim, Ab-dul Ghani, Surakat, Yusuf Alhaji, Balogun, Abdullateef Oluwagbemiga, Shuaibu, Aliyu Nuhu, Garba, Aliyu, Kumar, Ganesh, Abdulkarim, Muhammad, Sahalu, Yusra, Mohammed, Tanko Yahaya, Abdulkadir, Bashir Abubakar, Abba, Abdallah Alkali, Iliyasu Kakumi, Nana Aliyu, Mamman, Hussaini, Ridwan, Salahudeen |
| Předmět: |
|
| Zdroj: |
Proceedings of International Conference on Studies in Engineering, Science, & Technology; 2023, Vol. 1, p96-107, 12p |
| Abstrakt: |
Diabetes is a major public health concern affecting millions of individuals worldwide. Early and accurate diagnosis is crucial for effective management and prevention of complications. In recent years, deep neural networks (DNNs) have shown great promise in improving diabetes classification accuracy due to their ability to capture complex patterns in the data. However, several challenges must be addressed for optimal performance of DNNs, including data preprocessing, data imbalance, and hyperparameter tuning. In this research article, a comprehensive approach to enhancing diabetes classification using DNNs is proposed. First, data preprocess using feature scaling, dimensionality reduction, and missing value imputation techniques to improve the quality of the input data. Next, the issue of data imbalance is addressed using oversampling and undersampling techniques to balance the classes. Finally, hyperparameter tuning was performed using a combination of grid search and random search to identify the optimal set of hyperparameters for each DNN model. The performance of proposed approach was evaluated on a publicly available dataset of 768 patients and compare the performance of our DNN-based models with traditional machine learning algorithms. The obtained results showed that the proposed approach significantly improves diabetes classification accuracy compared to traditional machine learning algorithms. Specifically, an accuracy 90%, precision of 85%, F1 score of 82%, recall of 83%, and area under the ROC curve (AUC) of 94% were achieved. The findings suggest that addressing data preprocessing, data imbalance, and hyperparameter tuning are essential for optimal performance of DNNs in diabetes classification. The proposed approach can potentially lead to earlier and more accurate diagnosis of diabetes, enabling more effective treatment and management of the disease. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
