Diabetes can be divided into 4 types, with type 2 diabetes being the most common type in Thai people. Therefore, this research focuses on screening normal people and patients with type 2 diabetes, which is caused by the body not producing enough insulin. As a result, the process of absorbing blood sugar into energy in the cells in the body is abnormal or does not work at full efficiency. The result is There is a large amount of sugar accumulated in the blood. The objectives of the project are 1) to study the application of NIR in measuring diabetic patients and normal people 2) to determine the efficiency of data classification with the Support Vector Machine (SVM) algorithm. Using an infrared spectroscopy sensor (Near Infrared Spectroscopy), there were 35 normal volunteers and 45 people with type 2 diabetes in Narathiwat Province. The experiment was divided into 4 items: 1) testing the sensor, 2) designing the sensor box, 3) selecting the appropriate wavelength range, and 4) selecting the data normalization method. (Normalization) From the experimental results, it was found that using black, opaque boxes gave the highest efficiency. The frequency ranges suitable for classification are 410, 460, 535, 610, and 730 nm. In addition, the min-max data normalization method provides better accuracy than z-score and has the highest classification accuracy. at 90 percent
การแปล กรุณารอสักครู่..
