To establish a pre-test probability model of coronary heart disease (CHD) combined with cardiopulmonary exercise test (CPET) indexes and to compare the clinical effectiveness with Duke clinical score (DCS) and updated Diamond-Forrester model (UDFM), thus further explore the predictive value. 342 cases were used to establish the prediction model equation and another 80 cases were used to verify the effectiveness. The patients were divided into CHD group (n = 157) and non-CHD group (n = 185) according to coronary artery stenosis degree >50% or not. Combining DCS and UDFM as reference models with CPET indexes, a multivariate logistic regression model was established. The area under the ROC curve of the three models were calculated to compare the predictive effectiveness. There were significant differences in gender, chest pain type, myocardial infarction history, hypertension history, smoking, pathological Q wave and ST-T change between two groups (P < 0.01), as well as age, LVEF, heart rate at anaerobic domain, peak oxygen uptake in kilograms of body weight, percentage of peak oxygen uptake to the predicted value, the oxygen uptake efficiency slope and carbon dioxide ventilation equivalent slope (P < 0.05). Multivariate analysis showed gender, age, chest pain type, myocardial infarction history, hypertension history, smoking, pathological Q wave, ST-T change, and peak oxygen pulse were independent risk factors of CHD. The pre-test probability model of CHD combined with CPET indexes has good distinguish and calibrate ability, its prediction accuracy is slightly better than DCS and UDFM, which still needs to be verified externally in more samples.

建立冠心病（CHD）结合心肺运动试验（CPET）指标的预测概率模型，并与杜克临床评分（DCS）和更新的Diamond-Forrester模型（UDFM）比较临床效果，进一步探讨的预测值。通过342个案例建立了预测模型方程，并用80个案例验证了有效性。根据冠状动脉狭窄程度>50%将患者分为CHD组（n=157）和非CHD组（n=185）。结合DCS和UDFM作为参考模型，结合CPET指标，建立多元Logistic回归模型。计算三个模型的ROC曲线下面积以比较预测效果。两组患者性别、胸痛类型、心肌梗死史、高血压病史、吸烟、病理Q波、ST-T改变以及年龄、LVEF、无氧域心率、峰值摄氧量（公斤体重）、峰值摄氧量与预测值的百分比、摄氧效率斜率和二氧化碳通气当量斜率（P < 0.05）。多因素分析显示性别、年龄、胸痛类型、心肌梗死史、高血压病史、吸烟、病理性Q波、ST-T改变、峰值氧脉搏是冠心病的独立危险因素。结合CPET指标的CHD预测概率模型具有良好的区分和标定能力，其预测精度略优于DCS和UDFM，但仍需要外部在更多样本中进行验证。