K-Nearest Neighbors, decision trees and random forest for diabetes prediction

Sapriadi Sapriadi (Institut Kesehatan Helvetia, Medan, Indonesia)

Jurnal Mantik
Vol. 7 No. 4 (2024): February: Manajemen, Teknologi Informatika dan Komunikasi (Mantik)

Diabetes can be treated and managed by taking medications, such as insulin injections or oral drugs, that help lower the blood sugar level. However, medications alone are not enough to control diabetes. People with diabetes also need to monitor their blood sugar level regularly, follow a balanced diet, limit the intake of sugar and carbohydrates, and avoid alcohol and tobacco. They also need to check their feet, eyes, and kidneys for any signs of damage, and seek medical attention if they notice any problems. Machine learning algorithms can help predict diabetes by learning from historical data and finding patterns that are not easily detected by human experts. They can also handle high-dimensional and noisy data, such as medical images or genomic sequences, that are relevant for diabetes diagnosis. However, machine learning algorithms also have some limitations, such as requiring a lot of data and computational resources, being prone to overfitting or underfitting, and being difficult to interpret or explain. Therefore, we conclude that random forest and decision tree are the best machine learning algorithms for predicting diabetes, and we recommend using them for future research and applications.