發(fā)布時間：2018-09-17 18:59

【摘要】：本文對二值結(jié)局變量數(shù)據(jù),基于因果推斷理論,提出根據(jù)患者的生物標(biāo)記物進行最優(yōu)治療方案選擇的統(tǒng)計方法.這種方法基于CSTE(covariate-specific treatment effect)曲線和CSTE曲線的置信帶(SCB).CSTE曲線表示在給定生物標(biāo)記物(協(xié)變量)的條件下,處理組的條件平均處理效應(yīng).同時,CSTE曲線及其SCB可以被用于對特定的治療方案選擇適宜的患者.本文利用B-樣條方法估計CSTE曲線及其CSB,并推導(dǎo)了其近似大樣本性質(zhì).本文還通過模擬比較研究了CSTE曲線的置信帶的有限樣本性質(zhì),并闡述了CSTE曲線及其置信帶在真實數(shù)據(jù)中如何選擇最優(yōu)治療方案.
[Abstract]:Based on the data of binary outcome variables and the theory of causality inference, this paper proposes a statistical method to select the optimal treatment scheme according to the biomarkers of the patients. This method is based on the confidence band (SCB). CSTE curve of CSTE (covariate-specific treatment effect) curve and CSTE curve) to represent the conditional average processing effect of the treatment group under the condition of given biomarkers (covariables). At the same time, the CSTE curve and its SCB can be used to select suitable patients for a specific treatment regimen. In this paper, the CSTE curve and its CSB, are estimated by using the B-spline method and its approximate large sample properties are derived. In this paper, we also study the finite sample properties of the confidence band of CSTE curve by means of simulation and comparison, and explain how to select the optimal treatment scheme of the CSTE curve and its confidence band in the real data.
【作者單位】： 中北大學(xué)理學(xué)院;Department
【分類號】：O212.1
，

本文編號：2246807