Skip to Main Content
Allergic rhinitis is a prevalent disease throughout the world. Electrodermal screening devices (EDSD) are devices that can measure the electrical properties of acupuncture points. This paper performs a series of experiments based on machine learning algorithms to study the feasibility of utilizing EDSD to diagnose allergic rhinitis. The experimental result shows that, to assess the presence of allergic rhinitis, using the k-nearest neighbor classification algorithm, the accuracy can achieve 93.26%, and using the support vector machine classification algorithm, the average accuracy can achieve 97.78%. The experimental result also shows that using, respectively, the k-means clustering algorithm and the Ward's hierarchical clustering algorithm to cluster the data into three clusters, 87% of the data are consistently clustered. The average total symptom scores in these three clusters are also very consistent. Based on the 87% consistently clustered data, using the support vector machine algorithm to assess the severity (mild and moderate/severe) of allergic rhinitis, the average accuracy can achieve 99.57%. In particular, the experimental result also shows that the disordered EDSD values at acupuncture points of spleen meridian and liver meridian coincides with the clinic experiences of standard traditional Chinese medicine.