A dynamic model using k-NN algorithm for predicting diabetes and breast cancer.

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Computational Biology

Background:

• Diabetes and breast cancer are globally prevalent diseases with significant health impacts.
• Early detection is crucial for reducing mortality rates but remains challenging.
• Traditional machine learning models, including k-nearest neighbors (k-NN), often struggle with diverse medical datasets due to static parameters.

Purpose of the Study:

• To propose a novel dynamic k-nearest neighbors (k-NN) model for enhanced disease classification.
• To improve prediction accuracy by dynamically adjusting the 'k' value based on local data characteristics.
• To evaluate the proposed model's performance against state-of-the-art methods in medical data classification.

Main Methods:

• Development of a dynamic k-NN algorithm that adapts the 'k' value.
• Testing the proposed model on the PIMA Diabetes and Breast Cancer Wisconsin (BCW) datasets.
• Evaluation using metrics including accuracy, precision, recall, F1-score, and execution time.

Main Results:

• The dynamic k-NN model demonstrated improved performance on both PIMA Diabetes and BCW datasets.
• Specific metric results showed enhanced classification capabilities for these critical diseases.
• The proposed model outperformed several existing state-of-the-art machine learning models.

Conclusions:

• The dynamic k-NN model offers a more effective and efficient approach to medical data classification.
• This adaptive approach holds significant promise for improving early disease detection and diagnosis.
• Further research can explore the application of this model to a wider range of medical conditions.