Mathematical Model for Disease Prediction. A Quantitative Model for Prediction of Potential for Rocky Mountain Spotted Fever in the Western United States.

Abstract

A quantitative version of a mathematical model has been formulated for predicting the potential for Rocky Mountain Spotted Fever throughout the western United States. The model has been formulated using a training-and-test strategy. The input data included up to 140 possible geo-physical variables, along with an independent empiric estimation of disease potential. The latter independent estimate was taken to be the true classification for purposes of training and scoring the model. Prediction using a stepwise regression model with unweighted input variables yielded 94% correct for the high potential sites and 96% correct for the low potential sites during the training phase. The data base of observations was not sufficient to test the model with independent data. A 10% jacknifing procedure through the data base yielded 66% correct prediction for the high potential sites and 95% correct prediction for low potential sites. Using empirically weighted input variables did not materially improve predictions. Validation of the model is recommended using an entirely independent data set. (Modified author abstract)

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 1973
Accession Number
AD0769951

Entities

People

  • Donald A. B. Lindberg
  • Jian K. Chang
  • Lynn J. Walley
  • Samuel J. Dwyer Iii

Organizations

  • University of Missouri

Tags

DTIC Thesaurus Topics

  • Abstracts
  • Classification
  • Data Sets
  • Databases
  • Diseases And Disorders
  • Mathematical Models
  • Models
  • Mountains
  • Observation
  • Rocky Mountains
  • Tickborne Diseases
  • Training
  • United States
  • Validation

Readers

  • Computational Modeling and Simulation
  • Neurodegenerative Parkinson's Disease and Rickettsial Disease handbook, including the data level of dopamine, BC, neurons, and PD.