| Description |
There is a documented need for the ability to rapidly target high-risk individuals for diseases. Genetic testing and preventive medicine, for instance, could more efficiently direct their services to the community if they knew which portion of the population were the best candidates for testing or treatment. Since family history is a proven risk factor for many common diseases, genetic counselors have been able to estimate an individual's risk for a disease or for the presence of disease allele by pedigree analysis. For diseases with genetic component, an individual's risk for predisposing to a genetic condition can be estimated based on the familial pattern of the diseases. However pedigree analysis for risk estimation is a time-consuming, resource-intensive, laborious process impeded by the large amounts of data required, problems in collecting family history data, and the complexity of the Bayesian probability calculations used. The goal of this project was to create a mechanism to streamline the risk estimation process by automating data collection, manipulation, and risk calculation. This project was possible due to the creation of the Intermountain Health Care/Genetic Research Family Health Database, a large collection of genealogical records and clinical discharge diagnoses. Using this resource, the Pedigree Processing Program (PPP) was constructed to automate the process of pedigree construction, gathering the disease records for an individual's family members and performing the rick calculations for the probands of interest. Asthma and osteoporosis were selected as the diseases to study. The focus of the study was the construction of PPP and a statistical analysis of the utility and value PPP. The probabilities created by PPP were assessed in terms of their accuracy, reliability, and discriminatory power. Although the goal of creating an efficient mechanism for automating risk calculations was achieved, the data samples used were not able to completely validate the hypotheses posted in this case-control study. Reliability of the risk probabilities could not be measured due to lack of clinical data available and the lack of knowledge on the genetic characteristics of the diseases selected. However, discriminatory power was observed for some of the matched samples, and sufficient accuracy of discrimination based on the actual affection status of the proband was achieved. It could not be statistically determined which familial variable or risk probability variable best differentiated affected probands from their unaffected matched controls. Unfortunately , some of the positive results of this analysis could be explained by a bias in the ascertainment of the data. |
