A Study of Estimation Accuracy When Using a Logistic Model for Prediction of Impact Acceleration Injury.
DESMATICS INC STATE COLLEGE PA
Pagination or Media Count:
This report addresses the topic of estimation accuracy in the development of an empirically-based logistic model for predicting impact acceleration injury. Two items of central interest are the degree of accuracy which may be expected for predictions derived from a model and the sensitivity of such predictions to sample size. A Monte Carlo simulation study was undertaken to provide information relating accuracy to sample size for selected model configurations. Two specific sets of model parameters were considered, Monte Carlo samples of various sizes were generated for each, and the accuracy of the resultant predictions were evaluated with respect to the true probabilities. Based on this Monte Carlo study, it appears that the procedure used for estimating the coefficients of the logistic model works well. This procedure provides convergence to the correct coefficient values and true probabilities as the sample size increases. Thus, if injury probability as a function of head dynamic response variables may be approximated by a logistic function, the estimation procedure described in this report will yield a satisfactory approximation to that function. Author
- Medicine and Medical Research