Data used for the diagnosis of thyroid cancer patients include Thyroid Stimulating hormone (TSH), Thyroglobulin (Tg), Tg Antibodies (TgAb), and pathology information. However, existing thyroid cancer-related studies using machine learning tended to use a single type of data in model. Therefore, this study proposes a new model and method that can use all type of data together. The proposed method uses an autoencoder model composed of multiple encoders to extract comprehensive features from three types of hormone related data and pathology data and connects them to a classification layer to train them to be classified. As a result of the experiment, the proposed model using both types of hormonal and pathological data showed a performance improvement of up to about 58.1 times compared to the model using only a single type of data.
ACKNOWLEDGMENT This research was supported by BK21 FOUR program of the National Research Foundation of Korea funded by the Ministry of Education(NRF5199991014091), Institute for Information communications Technology Promotion(IITP) grant funded by the Korea government (MSIP) (No. S2022A068600023), and the Ajou University research fund.
