Rojas Jimena, Fernandez Itziar, Pastor J Carlos, Garcia-Gutierrez Maria-Teresa, Sanabria Rosa-Maria, Brion Maria, Sobrino Beatriz, Manzanas Lucia, Giraldo Antonio, Rodriguez-de la Rua Enrique, Carracedo Angel
Department of Ophthalmology, University of Valladolid, IOBA (Institute for Research in Ophthalmobiology), Valladolid, Spain.
Invest Ophthalmol Vis Sci. 2009 May;50(5):2384-90. doi: 10.1167/iovs.08-2670. Epub 2008 Dec 20.
Machine learning techniques were used to identify which of 14 algorithms best predicts the genetic risk for development of proliferative vitreoretinopathy (PVR) in patients who are experiencing primary rhegmatogenous retinal detachment (RD).
Data from a total of 196 single nucleotide polymorphisms in 30 candidate genes were used. The genotypic profile of 138 patients with PVR following primary rhegmatogenous RD and 312 patients without PVR RD were analyzed. Machine learning techniques were used to develop statistical predictive models. Fourteen models were assessed. Their reproducibility was evaluated by an internal cross-validation method.
The three best predictive models were the lineal kernel based on the Support Vector Machine (SMV), the radial kernel based on the SVM, and the Random Forest. Accuracy values were 78.4%, 70.3%, and 69.3%, respectively. The more accurate, although complex, algorithm uses 42 SNPs, whereas the simpler one uses only two SNPs, which makes them more suitable for routine diagnostic work. The radial kernel based on SVM uses 10 SNPs. The best individually predictor marker was rs2229094 in the tumor necrosis factor locus.
Genetic variables may be useful to predict the likelihood of the development of PVR. The predictive capabilities of these models are as good as those observed with clinical approaches. These results need external validation to estimate the true predictive capability and select the most appropriate ones for clinical use.
运用机器学习技术，确定14种算法中哪一种能最佳预测原发性孔源性视网膜脱离（RD）患者发生增殖性玻璃体视网膜病变（PVR）的遗传风险。
使用来自30个候选基因中总共196个单核苷酸多态性的数据。分析了138例原发性孔源性RD后发生PVR的患者和312例未发生PVR的RD患者的基因分型概况。运用机器学习技术建立统计预测模型。评估了14个模型。通过内部交叉验证方法评估其可重复性。
三个最佳预测模型分别是基于支持向量机（SMV）的线性核、基于支持向量机的径向核和随机森林。准确率分别为78.4%、70.3%和69.3%。更准确（尽管更复杂）的算法使用42个单核苷酸多态性，而更简单的算法仅使用两个单核苷酸多态性，这使得它们更适合常规诊断工作。基于支持向量机的径向核使用10个单核苷酸多态性。最佳个体预测标记是肿瘤坏死因子基因座中的rs2229094。
遗传变量可能有助于预测PVR发生的可能性。这些模型的预测能力与临床方法所观察到的相当。这些结果需要外部验证，以估计真正的预测能力，并选择最适合临床使用的模型。
