Pharmocogenetics, Clinical Avatars and Predictions of Personalized Medicine |
|
|---|---|
|
|
|
| Launch in standalone player | |
| Air date: | Wednesday, November 04, 2009, 10:30:00 AM Time displayed is Eastern Time, Washington DC Local |
| Category: | Special |
| Description: | NLM Informatics Lecture Series
Dr. Tonellato presents a general approach, mathematical model and computational method to predict clinical efficacy of genetic discoveries using ‘clinical avatars’ to conduct simulations of the effect of genotypes on risk, diagnosis and treatment. Clinical avatars are individual medical data records produced from a stochastic model and statistical parameters developed to reflect actual patient populations. The approach is used to detect differences between predictions of two warfarin dosing prediction algorithms applied to several representative patient populations (US general, African American, Asian). Clinical variables (clinical, prescription, and genetic) used in the model were derived from examination of published warfarin prediction and decision support algorithms. Clinical avatars are then produced with variables and population means, variances and dependencies consistent with those found in the literature. Simulations demonstrate strengths and weaknesses of the dosing algorithms depend on population characteristics, size and genetic frequencies. All modeling, computations and simulations are conducted on our cloud computing environment. Examples of the strategy, methods, and simulations in other areas of personalized medicine such as cancer risk prediction will be discussed. Acrobat Slides |
| Author: | Peter J. Tonellato, Ph.D., Senior Research Scientist, Center for Biomedical Informatics, Harvard Medical School |
| Runtime: | 90 minutes |
| CIT File ID: | 15413 |
| CIT Live ID: | 8136 |
| Permanent link: | http://videocast.nih.gov/launch.asp?15413 |