OBJECTIVE We proposed a 40-Hz auditory steady-state response (ASSR) automatic detection method, and studied the prognosis of comatose patients by combining the 40-Hz ASSR detection results of multiple paradigms of auditory stimulation. METHODS The 40-Hz ASSR elicitation experiments were carried out on 32 comatose patients, with the detection results used as prognosis predictors. To achieve automatic detection, the detection was modeled as a binary hypothesis test for a sinusoidal waveform with unknown amplitude and phase, based on the generalized likelihood ratio test (GLRT). The patients were followed up for 6 months, and each patient's outcome was classified as either favorable outcome (severe disability, moderate disability or good recovery) or unfavorable outcome (vegetative state/unresponsive wakefulness syndrome or death) according to the Glasgow outcome scale (GOS). The performance of the prognosis predictors was assessed using the area under the receiver operating characteristic curve (AUC-ROC). RESULTS The largest AUC in univariate analysis involving a single stimulation paradigm was 0.849, while the AUC obtained by combining multiple predictors was increased to 0.966. CONCLUSIONS For comatose patients, the absence of 40-Hz ASSR in multiple stimulation paradigms may indicate an unfavorable prognosis. Furthermore, the combination of multiple auditory stimulation paradigms may increase the outcome prediction accuracy. SIGNIFICANCE The combination of multi-paradigm 40-Hz ASSR automatic detection results may provide a feasible automatic outcome prediction method for comatose patients.

中文翻译：

---

40Hz听觉稳态反应自动检测方法及其在昏迷患者预后中的应用

目的提出一种40-Hz听觉稳态响应（ASSR）自动检测方法，结合多种听觉刺激范式的40-Hz ASSR检测结果，研究昏迷患者的预后。方法对32例昏迷患者进行40Hz ASSR诱发实验，检测结果作为预后预测指标。为了实现自动检测，基于广义似然比检验 (GLRT)，检测被建模为具有未知幅度和相位的正弦波形的二元假设检验。对患者进行了 6 个月的随访，每个患者的结果被归类为有利结果（严重残疾、根据格拉斯哥结果量表 (GOS)，中度残疾或恢复良好）或不利结果（植物人状态/无反应觉醒综合征或死亡）。使用受试者工作特征曲线下面积 (AUC-ROC) 评估预后预测因子的性能。结果 涉及单一刺激范式的单变量分析中的最大 AUC 为 0.849，而通过组合多个预测因子获得的 AUC 增加至 0.966。结论 对于昏迷患者，多重刺激模式中 40 Hz ASSR 的缺失可能表明预后不良。此外，多种听觉刺激范式的组合可以提高结果预测的准确性。