Classification of a biomedical dataset to predict human diseases

Objective: To predict whether the patient has a particular disease or not based on the features extracted from the patient's eye image

Dataset Description:

1. The dataset contains 19 features which are extracted from an image set to predict whether that image contains the signs of a disease or not
2. Each row corresponds to a patient's eye image
3. Each column corresponds to a feature (Total 19 columns for 19 features)

Feature Description (19 features)

Attribute	Description
0	The quality assessment
1	The result of pre-screening.
2 - 7	The results of Macula detection.
8 - 15	The results of Macula detection for exudates.
16	The Euclidean distance.
17	The diameter of the optic disc.
18	The AM/FM-based classification

Number of training and testing samples

• 920 training samples
• 231 testing samples

Label description

• 1 corresponds to having a disease
• 0 corresponds to not having a disease

Results:

Classification Algorithm	Accuracy
k-nearest neighbors	67.82%
Support Vector Machine	69.56%
Naive Bayes	56.04%
Decision Tree	63.47%
Random Forest	72.17%
Logistic Regression	75.21%

Logistic regression turns out to provide the highest accuracy to predict whether the patient has the disease or not i. e. 75.21%