Current Research at ISERP
Hierarchical Bayes Methods for Serial Dilution Assays
by Andrew Gelman (Statistics)
Epidemiologic studies to examine relations between exposure to environmental agents and disease play a large role in the struggle to improve public health. The reliability and utility of data from these studies depends to a large extent upon the quality of the measure of exposure to the agents of interest. In the case of environmental agents such as outdoor air pollutants, heavy metals, pesticides, and bioaerosols, the difficulties associated with exposure monitoring can place severe limitations on otherwise well-designed studies.
In serial dilution assays, as for biological measurement in general, detection limits are a major problem. A procedure that gives more accurate estimates at low concentration, thus extending the usual Â"detection limits,Â" would be significant for public health. In particular, low concentrations of allergens potentially have important effects on childhood asthma. More generally, many compounds (such as benzene and indoor allergens) in the environment are hypothesized to affect health adversely (cancer and asthma) in low concentrations.
A common design for estimating the concentrations of compounds in biological samples is the serial dilution assay, in which measurements are taken at several different dilutions of a sample. The reason for this setup is that the concentration in an assay is quantified by an automated optical reading of a color change, and there is a limited range of concentrations for which the color change is informative: at low values, the color change is imperceptible, and at high values, the color saturates. Thus, several dilutions give several opportunities for an accurate measurement.
Bayesian inference is a statistical approach to estimating groups of parameters in the context of multiple sources of uncertainty. This project involves Bayesian inference for serial dilution assays in order to improve the accuracy of estimates. Investigators are developing this approach in the context of estimating the concentrations of compounds that are related to asthma and allergies, an area where dilution assays are routinely used and where more improved estimates, especially for concentrations that are conventionally Â"below detection limits,Â" could lead to improvements in scientific understanding as well as make more efficient use of current and future lab data.
See Also
- Research grants undertaken by Andrew Gelman
- Featured publications by Andrew Gelman
Funded by National Institutes of Health »
