The student team of the Graduate Institute of Data Science win second place and Innovative Award in the AI CUP

Mentored by Associate Professor Yung-Chun Chang from the Graduate Institute of Data Science TMU students, Sheng-Jie Lin; Min-Chen Chen; and Chin-Ju Chen, outperformed more than 360 other participating teams and won the second place and the Innovative Award in this year’s AI CUP organized by the Ministry of Education.

Group photo of team members (from left): Min-Chen Chen, Sheng-Jie Lin, Yung-Chun Chang (associate professor) and Chin-Ju Chen

Under the theme of Medical Information Decision-making and Conversation Analysis Competition, AI CUP 2021 was an intense competition that promotes the development of solutions to medical-related issues using data analysis, model prediction and artificial intelligence.

The TMU team adopted Natural Language Processing (NLP) technology to process the content of the conversations. The highlight of this technology is that the dialogues of medical professionals as well as the dialogues of patients and family members are retrieved separately during the data pre-processing; hence, the text can be processed in a more detailed manner, and more characteristic data can be obtained.

The ColumnTransformer and K-fold cross validation are used in the “decision-making prediction and risk assessment task”, and then Ensemble Learning is used to generate the prediction results during model building. In the “Question and Answer Task between Medical Professionals and Patients”, the TMU team has developed the “STSBERT” algorithm, which can quickly filter and select the key sentences in the text and solve the issue of poor model prediction due to excessive text.

Led by Associate Professor Yung-Chun Chang, the team differentiated themselves from the other participating teams by proposing the methods and creative highlights, which can effectively improve the predictive ability of the model and present the most suitable one by compiling many experiments. The results of this competition can also be applied to other fields, benefiting the development of many more new technologies applicable to clinical practice in the future and providing health professionals with timely assistance to their decision-making .