발간년도 : [2022]
| 논문정보 |
|
| 논문명(한글) |
[Vol.17, No.5] Development of TKM School-Specific Prescription Recommendation Model Based on Word Embedding: Chronic Respiratory Disease |
|
| 논문투고자 |
Sang-Jun Yea, Sang-Hun Lee |
|
| 논문내용 |
In traditional medicine, the process of selecting appropriate prescriptions for the patients requires a complex clinical decision-making process, and a number of studies are being conducted to support these decision-making process. However, in previous studies, algorithmic studies reflecting the characteristics of each TKM (Traditional Korean Medicine) school are not carried out. Therefore, in this paper, we proposed a machine learning model which is divided into an input layer, an embedding layer, a classifier layer, and an output layer to recommend prescriptions for each TKM school. Also, experts from 3 TKM schools participated in building 480 machine learning data for chronic respiratory diseases. The performance of the proposed machine learning model was measured by the 5 folds cross validation method with the accuracy of recommending five prescriptions. For symptoms embedding, the model applied fastText encoder showed the best performance, and for the classifier, the model using SVM was found to have the best performance. In addition, the accuracy of each TKM school was the highest at 99.29% for the Sanghan-Geumgwe medical school and the average of the three schools was 85.42%. This study is meaningful in that is showed a novel approach to developing a machine learning model which is capable of recommending prescriptions for each TKM school. |
|
| 첨부논문 |
|
|
|
|
|
자료실 > 논문지 
