在職業球賽中，碰撞和衝突是不可避免的，這很容易導致創傷性腦損 傷，甚至危及生命。這些腦震盪症狀可能會嚴重影響運動員的職業生涯 和發展，而後遺症可能導致永久性損傷。因此，一種即時且有效的測試 或檢測方法是必要的工具。過去大多數的測試方法相對主觀，通常需要 運動員自行報告，這可能易於操縱。過去可以使用視覺誘發電位（VEP） 測試人腦的視覺功能，但需要一個安靜和無干擾的環境來收集數據。腦 震盪患者通常在某些腦部區域有頭部受傷。隨著虛擬現實（VR）和機器 學習（ML）的快速發展，我們利用VR 頭戴式設備創造出一個沉浸式環 境，這個環境可以創造一個無干擾的封閉環境，以更可靠地從患者收集 數據。多模態融合方法可以檢測腦震盪症狀並對應不同的神經功能進行 全面評估，使用腦電圖（EEG）和眼動追蹤傳感器。最近，機器學習和 深度學習方法在發現腦震盪症狀方面應用較少，因此我們進一步研究這 些先進技術。我們提出了一種新穎的腦震盪檢測系統，結果在我們的模 擬數據集中實現了0.95 的準確率。;In professional ball games, collisions and conflicts are inevitable, which can easily cause traumatic brain injuries and even life-threatening injuries. These concussion symptoms can seriously impact a player’s career and development, and the sequelae can cause permanent damage. Therefore, an immediate and effective testing or detection method is a necessary tool. Most of the testing methods in the past were relatively subjective, and they usually need a self-report for players themselves, which may cause easily be manipulated. In the past, a visual evoked potential (VEP) can be used to test the visual function of the human brain but needed a silent and interference-free environment to collect data. Concussion patients often have head injuries in some brain sections. With the rapid development of Virtual Reality (VR) and Machine Learning (ML), we utilize a VR headset to create an immersive environment that creates an enclosed environment that is interference-free in order to collect data more reliably from patients. Multi-modal fusion methods can detect concussion symptoms and make a comprehensive assessment for different brain sections corresponding with different neurological functions with EEG and eye-tracking sensors. Recently, ML and DL methods are less applied to find concussion symptoms, then we further investigate these with advanced techniques. We propose a novel concussion detection system and the result can achieve an accuracy of 0.95 in our simulation dataset.
