Semi-automated initialization of simulations: an application to healthcare. (April 2016)

• Record Type: Journal Article
• Authors: Padilla, Jose J; Diallo, Saikou Y; Kavak, Hamdi; Sahin, Olcay; Sokolowski, John A; Gore, Ross J
• Is Part Of: Journal of defense modeling and simulation. Volume 13:Number 2(2016:Apr.)
• Page Start: 171
• Page End: 182
• Publication Date: 2016-04
• Subjects: Simulation initialization -- data-mining -- information extraction -- social media; Military art and science -- Computer simulation -- Periodicals; 355.0011305
• DOI: 10.1177/1548512914565503
• Languages: English
• ISSNs: 1548-5129

Abstract

As the number of variables and entities included in a simulation model increase, it becomes more difficult to initialize due to (a) the increasing number of input variables that are required and (b) the difficulty in finding, retrieving, and assigning the initial values of the input variables, especially in Human Social Cultural Behavior Modeling. As a result, the initialization process is generally more time consuming and error prone which motivates the need for semi-automated approaches wherever possible. In this paper, we propose a semi-automated approach for initializing input variables that are challenging to quantify and require additional processing to be assigned their initial values. We apply this approach to initialize a healthcare simulation using data from sources such as the US Census Bureau, County Health Rankings, and Twitter. Results show that the approach works well when we can find variables with existing values even for large input data sets (over 50 variables). However, additional work is required to determine whether the values assigned using this approach yield more accurate simulation results.