Abstract:
|
Clinical data is often characterized by some degree of inhomogeneity. This is typically the case in meta-analysis, where we combine results from different studies and in analysis of large scale data, where the data contains certain degrees of inhomogeneity, due to sample size. In this paper, we present a general methodology to deal with such big data. More precisely, we propose a maximin effects for treatment decision making, to aggregate information over patients of all subpopulations. The estimator of the proposed maximin effects can be obtained by solving a quadratically constrained linear programming (QCLP), which can be efficiently computed by interior-point methods. Consistency and asymptotic normality of the estimator is established. In addition, we study the limiting distribution of the estimated value function for the obtained "maximin treatment decision". Numerical examples show our maximin estimator performed more reliably than the "pooled effects" estimator and the "simple average" estimator.
|