Hierarchical Bayes Methods for Serial Dilution Assays

描述(申请人提供)：连续稀释是测量生物样品中未知化合物浓度时广泛使用的关键步骤。在这些分析中，低于检测限的测量是一个长期存在的问题。我们将使用分层贝叶斯推断-一种在存在不确定性的情况下估计参数组的统计方法-来改进连续稀释分析的估计，从而允许估计以前被确定为“低于检测限值”的浓度。我们将开发一个使用开源软件的程序，以便来自世界各地的实验室的研究人员可以评估和使用新方法。我们将在已知条件下进行实验室验证研究，评估新方法。我们将立即将这种方法应用于从有哮喘风险的儿童家中收集的粉尘样本中的过敏原的实验室研究。通过扩大检测范围，改进的估计程序将在儿童哮喘的研究中特别有用，在儿童哮喘的研究中，即使是非常低的过敏原浓度也被假设为对健康有不利影响。我们还将进行一系列实验和数据分析，以扩展该模型，以考虑到由于样本污染而导致的校准曲线的变化，这是环境样本研究和更广泛的生物测试中的一个常见问题。我们将探索使用我们的估计程序进行更有效的系列稀释实验设计的可能性。我们的方法将在我们对过敏原和哮喘的实验室研究的背景下开发；然而，我们预计它将更普遍地适用于许多不同生物环境中的系列稀释分析。对低暴露水平的测量在一般公共卫生问题中是至关重要的，因此，该项目对不确定源进行了明确的建模，从而在低水平下产生了更准确的估计，这可能会导致更有效的一般生物测定。

DESCRIPTION (provided by applicant): Serial dilution is a crucial step that is widely used when measuring the concentrations of unknown compounds in biological samples. Measurements below detection limits are a persistent problem in these assays. We will use hierarchical Bayes inference-a statistical approach for estimating groups of parameters in the presence of uncertainty-to improve estimation for serial dilution assays, thus allowing estimation of concentrations that would previously have been identified as "below detection limits." We will develop a program using open-source software so that researchers from laboratories around the world can evaluate and use the new method. We will perform a laboratory validation study evaluating the new method under known conditions. We will immediately apply the methodology to laboratory studies of allergens in dust samples collected from the homes of children who are at risk for asthma. By extending limits of detection, the improved estimation procedure will be particularly helpful in the study of childhood asthma, where even very low allergen concentrations are hypothesized to have adverse health effects. We will also undertake a series of experiments and data analyses to extend the model to allow for changes in the calibration curve due to contamination of the samples, which is a common problem in the study of environmental samples, and in bioassays more generally. We will explore possibilities of more efficient designs of serial dilution experiments using our estimation procedure. Our method will be developed in the context of our laboratory studies of allergens and asthma; however, we anticipate it will be applicable much more generally to serial dilution assays in many different biological contexts. Measurement of low levels of exposure is critical in public health problems in general, hence this project, in which sources of uncertainty are explicitly modeled, leading to more precise estimates at low levels, will potentially lead to more effective bioassays generally.