人工智能与脑网络组融合助力意识障碍预后预测研究

Artificial intelligence and brain network research improve the prognostications of disorders of consciousness

意识障碍(DOC)患者的预后预测对临床治疗及患者家属都有重要意义。但目前的预后预测模型的准确度不高,敏感性和特异性均较低。本研究在基于病因、年龄和病程作为重要预测指标的基础上,融合使用基于脑功能磁共振影像的患者脑功能网络特征,利用人工智能和机器学习算法,研发出了一个预测意识障碍患者能否苏醒的计算模型,准确率达到了88%。

Although some indicators and models have been proposed to prognosticate disorders of consciousness(DOC),any single method when used alone carries a high risk of false prediction.This study aimd to develop a multidomain prognostic model that combined functional brain networks with clinical characteristics to predict one year outcomes at the single-patient level.The model discriminate patients who would later recover consciousness from those who would not with an accuracy of around 88%.