基于神经网络算法与液体蛋白芯片指纹图谱建立矽肺早期诊断模型

Diagnostic Prediction of Early Silicosis Patients Using Neural Network and MALDI-TOF-MS in Serum

应用磁珠分选和基质辅助激光解吸电离飞行时间质谱(MALDI-TOF-MS)联合技术检测,获得了79例矽尘暴露人群和25例健康对照样本的血清蛋白质谱,从矽尘暴露人群与健康对照组中筛选出7个共有的蛋白质谱差异峰,结合人工神经网络(ANN)建立早期矽肺患者诊断模型,用建立的诊断模型进行盲法检测,该诊断模型的特异性为84.52%,敏感性为93.69%,准确率为91.35%。在矽尘暴露各组和健康对照组之间进一步建立了神经网络辨别模型,对无尘肺0期、无尘肺0+期和Ⅰ期矽肺组的正确区分率分别达到89.23%、94.20%和92.37%。

Serum of 79 workers exposed to silica and 25 healthy controls cases were determined by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry（MALDI-TOF-MS）.7 protein peaks were selected and used by artificial neural network（ANN） to establish a diagnostic model.A blinded test showed that accuracy,sensitivity and specificity were 91.35%,93.69%,and 84.52%,respectively.The diagnostic pattern was also established to distinguish each stage of silica-exposed population.The diagnostic pattern worked excellently with 89.23%,94.20% and 92.37% of accurate rate for classifying phase 0,phase 0＋,and phaseⅠof silicosis,respectively.