吞嚥困難是現代社會中非常重要的議題，常見於中風患者以及年長者。在許多研究中顯示，舌頭強度可以做為吞嚥功能的評估標準。本研究使用了舌壓力儀器作為評估舌頭強度的工具，並且使用表面肌電圖儀器收集喉部肌肉電數據，再將評估任務結合有趣的遊戲，提高使用者使用的意願以及刺激使用者有更好的表現。任務完成後系統會將舌壓數據以及肌肉電數據收集起來，我們透過Scoring Function的運算，量化出使用者的吞嚥動作表現品質。除了計算動作的品質分數之外，我們也從收集到的數據中提取出特徵，建立多種機器學習模型比較每個模型的分類成效，挑選出最佳的模型藉以正確的預測出使用者吞嚥功能的等級高低。透過這套評估系統，我們希望可以提供快速且準確的評估結果，使得在醫療方面，醫療人員在進行診斷以及復健訓練時，能有更加方便以及有效的工具。;Dysphagia is a very important issue in modern society, and it is common in stroke patients and the elderly. In many studies, it has been shown that tongue strength can be used as an evaluation criterion for swallowing function. In this study, a tongue pressure instrument was used as a tool to assess tongue strength, and a surface electromyography instrument was used to collect electrical data of larynx muscles, and then the assessment task was combined with interesting games to increase users′ willingness to use and stimulate users to have better performance. After the task is completed, the system collects tongue pressure data and muscle electrical data. We use the Scoring Function calculation to quantify the user′s swallowing performance quality. In addition to calculating the quality score of the motion, we also extract features from the collected data, build a variety of machine learning models to compare each model’s classification effectiveness and select the best model to correctly predict the level of the user′s swallowing function. Through this evaluation system, we hope to provide fast and accurate evaluation results, so that medical personnel can have more convenient and effective tools for diagnosis and rehabilitation training.
