论文部分内容阅读
使用近红外光谱分析方法测量培养后的胚胎培养液,结合偏最小二乘判别分析对胚胎发育潜能进行评价,鉴别具有妊娠能力与不具妊娠能力的胚胎。为了提高模型的判别能力,消除无信息变量对模型稳定性影响,分别采用基于蒙特卡罗的无信息变量消除法(MC-UVE)、竞争性自适应加权抽样法(CARS)与基于变量稳定性的竞争性自适应加权抽样法(SCARS),对光谱进行波长选择。结果表明,与采用全谱74%的正判率相比较,采用这3种波长选择方法,模型独立检验集的正判率分别提高至74.24%,77.12%与80.10%,建模使用变量数降至50以内。比较发现,SCARS的模型优化能力和稳定性均好于MC-UVE和CARS方法。采用近红外光谱结合化学计量学方法预测胚胎的发育潜能是可行的。
The embryo culture fluid was measured by near-infrared spectroscopy and combined with partial least-squares discriminant analysis to evaluate the embryonic development potential to identify the embryo with or without pregnancy. In order to improve the discriminant ability of the model and eliminate the influence of non-information variables on the stability of the model, Monte Carlo-based MC-UVE, CARS and variable- Competitive adaptive weighted sampling method (SCARS), the wavelength of the spectrum selection. The results show that compared with the 74% correct rate of full spectrum, the correct rate of the model independent test set is increased to 74.24%, 77.12% and 80.10% respectively by using the three wavelength selection methods. The number of modeling variables To 50 or less. Compared with MC-UVE and CARS, SCARS has better model optimization ability and stability. Near infrared spectroscopy combined with chemometric methods to predict embryonic development potential is feasible.