Lael Gatewood - US grants

A simulation facility for stochastic micropopulation models at the University of Minnesota utilizes a VAZ 11/750 and VMS to run a locally-developed software system called SUMMERS (Simulation Utilities and Monte Carlo Methods for Epidemiological Research Studies). Since SUMMERS is highly modularized, it can be adapted to the simulation of different populations and models of different diseases by changing a limited number of program modules and variable subsets. Various program modules operate the user interfaces, the simulation process, the control functions, the report generation, and the analytic interpretation. Extensive software is being developed for sensitivity analyses of the multivariate parameters which are estimated stochastically. Hardware and software are assembled and integrated to enable convenient, research-oriented studies of stochastic population models. Current model development and research focuses on models of infectious disease epidemics, chronic cardiovascular disease, genetic ascertainment of diabetes, and data distributions from the clinical laboratory. The results of these simulations are of use in analyzing mechanisms of disease incidence, prevalence and case-finding, as well as in exploring strategies for prevention or limitation of spread. These studies and the outcomes of collaborative studies help test the applicability and friendliness of the facility offerings. It will also train new users, disseminate materials concerning the research and facilities available, and provide modeling services for remote users. During the first three years of renewal, different transfer modes will be developed for external users, ranging from multi-user, multi-function programming systems running on super micro- or mini-computers, to single user, single function systems dedicated to a particular version or model type, and to smaller microcomputer versions suitable for demonstration in the classroom or at seminars. Network facilities will also be investigated to link the community of SUMMERS users to other BRT simulation facilities and to each other.

The National Simulation Resource for Stochastic Micropopulation Models provides an environment for Monte Carlo modeling of structured micropopulations. Core research in simulation methodologies enhances collaborative development of categorical models in the areas of infectious, chronic, and genetic disease as well as cognitive processing. Commonalities at the conceptual, coding and implementation levels allow easy maintenance, sharing and reuse of the developed software. Current areas of categorical research examine the epidemic spread of HIV/AIDS, influenza and an animal model of herpes virus; population interventions against heart attack, stroke and other cardiovascular diseases; familial transmission of diabetes, epilepsy, schizophrenia and lung cancer; and artificial neural networks modeling visual processing. The simulation shell can also generate structured populations with given genetic, demographic and physiological characteristics to test analytic systems. An outreach program supports external users with dissemination, training and service. The goal of the Resource is to provide an integrated simulation environment to conduct biomedical research, from the creation of a model to the analysis of simulation results, and allows investigators to incorporate experimental data and evaluate hypotheses. A simulation shell termed SUMMERS, based on a single conceptual model, contains a library of VAX/VMS FORTRAN modules allowing the simulation of a generic population of individual members in interactive groups transferring between multiple activity states at discrete intervals. Researchers use keywords to modify existing categorical models or assemble program modules from the simulation shell and its templates to develop their own applications. A separate sensitivity analysis system assesses the relative importance and plausible range for model parameters given the outcome data, and can optimize parametric values to fit pre-specified outcomes. Prototype expert systems provide control frameworks to help build, implement, modify and analyze these SUMMERS-based models. New initiatives planned include incorporating these subsystems into the SUMMERS shell, adapting the SUMMERS code to C and UNIX-based computer systems, enhancing interactive and graphics-based use, and developing object-oriented structures within SUMMERS to represent social networks and other micropopulation features.