This paper describes two simulation models developed in the course of a research project undertak... more This paper describes two simulation models developed in the course of a research project undertaken for the North Carolina Department of Mental Health. The project performed a systems analysis of the alcoholism program currently being conducted by the Department in one of four regions in the state. One model is a micro simulation of individuals with problem drinking behavior and the resulting community actions in the handling of that behavior. This model is comprehensive in its detail; aspects of treatment, rehabilitation, program effectiveness, and cost are incorporated in order to provide extensive statistics used in the second model. The second model is a macro aggregative simulation which operates in a conversational mode and is used by management within the Department of Mental Health in their planning and budgeting process. The details of each model are discussed, results from the models are presented, and the initial reactions of the executives within the Department of Mental Health are summarized.
Abstract—This paper describes the use of validation methods in model building. We address issues ... more Abstract—This paper describes the use of validation methods in model building. We address issues associated with the increasing complexity of models that is in part a response to the growing popularity of Agent-Based Models (ABM), commonly used to study cognitive, natural, and social phenomena. The first section of this manuscript discusses model categories and attributes. The second section discusses the stages of validating a simulation model: verification, validity, and sensitivity analysis. The third section presents specific validation approaches, with an emphasis on six specific tests that are described in detail. The final section summarizes the goals of model validation and modeling.
OBJECTIVE Agent-based models (ABMs) have been developed to simulate epidemics including smallpox ... more OBJECTIVE Agent-based models (ABMs) have been developed to simulate epidemics including smallpox (Longini et al.) and pandemic flu (Ferguson et al.). The ABM approach is an effective method to assess the poten tial impact of interventions on disease spread (Fer guson et al.). Integrating the ABM approach with syndromic surveillance data provides potential benefit s such as permitting a realistic specification of som e critical model contact parameters, and permitting synthetic outbreaks to be generated with extremely fine resolution (e.g., age, gender, and address). T his would provide the ability to test various clusterin g detection algorithms – a key component of syndromic surveillance methods. RTI International (the Models of Infectious Disease Agent Study (MIDAS) informatics group) and NYC DOHMH (a premier syndromic surveillance research center) collaborated t o create a NYC-ABM of flu transmission. This poster describes implementation of several features requir d for accurate ...
... Philip C. Cooley, Ph.D.,' David N. Hamill, Ph.D., Susan J. Reade-Christopher, Ph.D., Els... more ... Philip C. Cooley, Ph.D.,' David N. Hamill, Ph.D., Susan J. Reade-Christopher, Ph.D., Elsa C. Liner, Ph.D., and Charles M. Van Der ... che e' stato sviluppato al fine di conoscere le correlazioni tra i parametri che caratterizzano I'epidemia e che includono: 1. I'uso comune di aghi e ...
Review of Quantitative Finance and Accounting, 2008
... 125 Huang, Roger D. 6 5.00 166 Chan, Louis KC 6 2.33 126 Schwert, G. William 6 5.00 167 Noe, ... more ... 125 Huang, Roger D. 6 5.00 166 Chan, Louis KC 6 2.33 126 Schwert, G. William 6 5.00 167 Noe, Thomas H. 6 2.33 127 Markowitz, Harry M. 6 4.83 168 Harris, Robert S. 6 2.25 169 Lease, Ronald C. 6 2.00 210 Kaufman, George G 5 2.67 292 JL Heck, PL Cooley 123 Page 7. ...
Information from 12 studies is combined to estimate the AIDS incubation distribution with greater... more Information from 12 studies is combined to estimate the AIDS incubation distribution with greater precision than is possible from a single study. The analysis uses a hierarchy of parametric models based on a four-parameter generalized F distribution. This general model contains four standard two-parameter distributions as special cases. The cases are the Weibull, gamma, log-logistic, lognormal distributions. These four special cases subsume three distinct asymptotic hazard behaviors. As time increases beyond the median of approximately 10 years, the hazard can increase to infinity (Weibull), can plateau at some constant level (gamma), or can decrease to zero (log-logistic and lognormal). The Weibull, gamma and 'log-logistic distributions' which repre
Background: During the 2009 H1N1 influenza epidemic, policy makers debated over whether, when, an... more Background: During the 2009 H1N1 influenza epidemic, policy makers debated over whether, when, and how long to close schools. While closing schools could have reduced influenza transmission thereby preventing cases, deaths, and health care costs, it may also have incurred substantial costs from increased childcare needs and lost productivity by teachers and other school employees. Methods: A combination of agent-based and Monte Carlo economic simulation modeling was used to determine the cost-benefit of closing schools (vs. not closing schools) for different durations (range: 1 to 8 weeks) and symptomatic case incidence triggers (range: 1 to 30) for the state of Pennsylvania during the 2009 H1N1 epidemic. Different scenarios varied the basic reproductive rate (R 0) from 1.2, 1.6, to 2.0 and used case-hospitalization and case-fatality rates from the 2009 epidemic. Additional analyses determined the cost per influenza case averted of implementing school closure. Results: For all scenarios explored, closing schools resulted in substantially higher net costs than not closing schools.
5,7 for the HIVNET Vaccine Preparedness Study Protocol Team Recent studies have reported on the u... more 5,7 for the HIVNET Vaccine Preparedness Study Protocol Team Recent studies have reported on the utility of audio computer-assisted self-interviewing (ACASI) in surveys of human immunodeficiency virus (HIV) risk behaviors that involve a single assessment. This paper reports the results of a test of ACASI within a longitudinal study of HIV risk behavior and infection. Study participants (gay men (n = 1,974) and injection drug users (n = 903)) were randomly assigned to either ACASI or intervieweradministered assessment at their second follow-up visit 12 months after baseline. Significantly more of the sexually active gay men assessed via ACASI reported having sexual partners who were HIV antibody positive (odds ratio = 1.36, 95% confidence interval: 1.08, 1.72), and a higher proportion reported unprotected receptive anal intercourse. Among injection drug users (IDUs), our hypothesis was partially supported. Significantly more IDUs assessed via ACASI reported using a needle after another person without cleaning it (odds ratio = 2.40, 95% confidence interval: 1.34, 4.30). ACASI-assessed IDUs reported similar rates of needle sharing and needle exchange use but a lower frequency of injection. Participants reported few problems using ACASI, and it was well accepted among members of both risk groups. Sixty percent of the participants felt that the ACASI elicited more honest responses than did interviewer-administered questionnaires. Together, these data are consistent with prior research findings and suggest that ACASI can enhance the quality of behavioral assessment and provide an acceptable method for collecting self-reports of HIV risk behavior in longitudinal studies and clinical trials of prevention interventions.
To evaluate the effect of the assumption of no long reporting delays on estimates of AIDS inciden... more To evaluate the effect of the assumption of no long reporting delays on estimates of AIDS incidence. Reported AIDS cases must be adjusted for reporting delays to estimate AIDS incidence. We compared the adjustments supplied with the Centers for Disease Control and Prevention (CDC) AIDS Public Information Data Set with a set of adjustments that differ with respect to CDC assumption of no long delays. Both methods assume that no changes in delay have occurred throughout the reporting period. Probability distributions of reporting delays were calculated from the delay adjustment weights supplied by CDC, and from an alternative method that estimates the probability of long delays from the surveillance data. AIDS incidence estimates from these two approaches were calculated and compared. The CDC adjustments assume that there will be no reporting delays longer than 61 months, whereas the alternative method estimates that 5.9% of case reports will be delayed longer than 61 months. The CDC adjustments result in lower estimates of AIDS incidence and a flattening of the epidemic. In addition to a 6.2% reduction in total estimated AIDS incidence, the CDC assumption changes the shape of the estimated epidemic. These may result in as much as 4-21% reductions in model estimates of HIV incidence and prevalence.
Computational and Mathematical Organization Theory, 2015
Many published influenza models treat each simulation day as a weekday and do not distinguish wee... more Many published influenza models treat each simulation day as a weekday and do not distinguish weekend days. Consequently, the weekend effect on influenza transmission has not been fully explored. To assess whether distinguishing between weekday and weekend transmissions in simulation models of flu-like infectious disease models matters, this study uses an agent-based model of the Chicago, Illinois metropolitan area. Our study assesses whether including weekend effects is offset by increases in weekend contact patterns and if implementing 3-day weekends dampens disease transmission enough to warrant its use as a containment strategy. Results indicate that ignoring weekend behaviors without incorporating increases in community-based non-school contacts (i.e., compensatory behaviors) causes the peak case incidence day to occur 7 days earlier and can reduce the peak attack rate by as much as 60 %. These results are sensitive to the proportion of symptomatic cases that are assumed to remain at home until they recover. The 3-day weekend intervention has interesting possibilities, but the benefits may only be effective for mild epidemics. However, a 3-day weekend for schools would be less detrimental to the educational process than sustained permanent closing because student and teacher contact is maintained throughout the epidemic period. Also, a 4-day school and work week may be more easily accommodated by many types of schools and businesses.
Maternal nutrition influences a child’s birthweight, which affects the child’s growth and subsequ... more Maternal nutrition influences a child’s birthweight, which affects the child’s growth and subsequent survival. However, the broad consequences of maternal undernutrition and the outcomes of interventions to improve maternal nutrition take years to manifest. To examine the long-term health outcomes of low birthweight infants in response to a maternal nutritional supplementation intervention without this obstacle, we developed the Forecasting Population Progress (FPOP) microsimulation model. The intervention we assessed was based on the findings of a published clinical trial outcome that reduced the incidence of low birthweight, a known cause of stunting. We implemented the “before intervention” and “after intervention” simulations and generated the difference in outcomes, using a spatially explicit synthetic baseline population of Indonesia generated from a microdata sample of the Indonesian 2010 census. We focused specifically on two provinces—Yogyakarta and Bali—which represent dif...
systems for accurate and cost-effective prediction of human health outcomes. Development of incre... more systems for accurate and cost-effective prediction of human health outcomes. Development of increasingly sophisticated in vitro tissue models has accelerated in recent years, driven by the recognition that conventional two-dimensional (2D) cell culture formats do not adequately recapitulate the 3D arrangements of cells and extracellular matrix of tissues and 1 Introduction 1.1 Organotypic cell cultures An intense area of research aims at improving drug development and toxicology assessment by shifting away from traditional animal studies and towards human relevant, in vitro
Measurements of drug use and other illicit or stigmatized behaviors are subject to nontrivial und... more Measurements of drug use and other illicit or stigmatized behaviors are subject to nontrivial underreporting biases. During in-person surveys, respondents are more likely to report such behaviors when interviewed using techniques that maximize interviewee privacy, e.g., use of paper SAQs and audio-CASI rather than questioning by human interviewers. Until recently, respondents in telephone surveys could not be offered similar privacy. A new technology, telephone audio computer-assisted self-interviewing (T-ACASI) overcomes this limitation of telephone surveys by allowing respondents to respond to a computer. A randomized experimental test of T-ACASI was embedded in the Urban Men's Health Study (UMHS). UMHS surveyed a probability sample of 2,881 men from four United States cities and who reported having sex with men. Respondents interviewed using T-ACASI reported a higher prevalence of drug use and drug-related behaviors than respondents interviewed by human interviewers. However, survey respondents were more likely to break off an interview when the interview was conducted by a T-ACASI computer rather than by a human interviewer.
We present an infectious disease model with the following characteristics: • Eight disjoint risk ... more We present an infectious disease model with the following characteristics: • Eight disjoint risk groups that are based on transmission via sexual and intravenous drug use in homosexual men, bisexual men, heterosexual men and women. • Contacts that follow from a homogeneous mixing assumption which takes into account risk group sizes and behavior. • A disease incubation function based on the results of Bachetti and Moss (1989). • No infections associated with hemophiliacs, blood product users, children under 13 years of age, and foreign borns. Simulation experiments were conducted to estimate critical model parameters such as the number of sexual and/or needle sharing contacts needed to explain observed AIDS cases and deaths due to HlV-related illnesses. These estimated parameters were used to make short-term national projections of the incidence of HIV and AIDS. I. INTRODUCTION This paper presents a simple model of the AIDS epidemic developed to explore some fundamental relationships...
This paper describes two simulation models developed in the course of a research project undertak... more This paper describes two simulation models developed in the course of a research project undertaken for the North Carolina Department of Mental Health. The project performed a systems analysis of the alcoholism program currently being conducted by the Department in one of four regions in the state. One model is a micro simulation of individuals with problem drinking behavior and the resulting community actions in the handling of that behavior. This model is comprehensive in its detail; aspects of treatment, rehabilitation, program effectiveness, and cost are incorporated in order to provide extensive statistics used in the second model. The second model is a macro aggregative simulation which operates in a conversational mode and is used by management within the Department of Mental Health in their planning and budgeting process. The details of each model are discussed, results from the models are presented, and the initial reactions of the executives within the Department of Mental Health are summarized.
Abstract—This paper describes the use of validation methods in model building. We address issues ... more Abstract—This paper describes the use of validation methods in model building. We address issues associated with the increasing complexity of models that is in part a response to the growing popularity of Agent-Based Models (ABM), commonly used to study cognitive, natural, and social phenomena. The first section of this manuscript discusses model categories and attributes. The second section discusses the stages of validating a simulation model: verification, validity, and sensitivity analysis. The third section presents specific validation approaches, with an emphasis on six specific tests that are described in detail. The final section summarizes the goals of model validation and modeling.
OBJECTIVE Agent-based models (ABMs) have been developed to simulate epidemics including smallpox ... more OBJECTIVE Agent-based models (ABMs) have been developed to simulate epidemics including smallpox (Longini et al.) and pandemic flu (Ferguson et al.). The ABM approach is an effective method to assess the poten tial impact of interventions on disease spread (Fer guson et al.). Integrating the ABM approach with syndromic surveillance data provides potential benefit s such as permitting a realistic specification of som e critical model contact parameters, and permitting synthetic outbreaks to be generated with extremely fine resolution (e.g., age, gender, and address). T his would provide the ability to test various clusterin g detection algorithms – a key component of syndromic surveillance methods. RTI International (the Models of Infectious Disease Agent Study (MIDAS) informatics group) and NYC DOHMH (a premier syndromic surveillance research center) collaborated t o create a NYC-ABM of flu transmission. This poster describes implementation of several features requir d for accurate ...
... Philip C. Cooley, Ph.D.,' David N. Hamill, Ph.D., Susan J. Reade-Christopher, Ph.D., Els... more ... Philip C. Cooley, Ph.D.,' David N. Hamill, Ph.D., Susan J. Reade-Christopher, Ph.D., Elsa C. Liner, Ph.D., and Charles M. Van Der ... che e' stato sviluppato al fine di conoscere le correlazioni tra i parametri che caratterizzano I'epidemia e che includono: 1. I'uso comune di aghi e ...
Review of Quantitative Finance and Accounting, 2008
... 125 Huang, Roger D. 6 5.00 166 Chan, Louis KC 6 2.33 126 Schwert, G. William 6 5.00 167 Noe, ... more ... 125 Huang, Roger D. 6 5.00 166 Chan, Louis KC 6 2.33 126 Schwert, G. William 6 5.00 167 Noe, Thomas H. 6 2.33 127 Markowitz, Harry M. 6 4.83 168 Harris, Robert S. 6 2.25 169 Lease, Ronald C. 6 2.00 210 Kaufman, George G 5 2.67 292 JL Heck, PL Cooley 123 Page 7. ...
Information from 12 studies is combined to estimate the AIDS incubation distribution with greater... more Information from 12 studies is combined to estimate the AIDS incubation distribution with greater precision than is possible from a single study. The analysis uses a hierarchy of parametric models based on a four-parameter generalized F distribution. This general model contains four standard two-parameter distributions as special cases. The cases are the Weibull, gamma, log-logistic, lognormal distributions. These four special cases subsume three distinct asymptotic hazard behaviors. As time increases beyond the median of approximately 10 years, the hazard can increase to infinity (Weibull), can plateau at some constant level (gamma), or can decrease to zero (log-logistic and lognormal). The Weibull, gamma and 'log-logistic distributions' which repre
Background: During the 2009 H1N1 influenza epidemic, policy makers debated over whether, when, an... more Background: During the 2009 H1N1 influenza epidemic, policy makers debated over whether, when, and how long to close schools. While closing schools could have reduced influenza transmission thereby preventing cases, deaths, and health care costs, it may also have incurred substantial costs from increased childcare needs and lost productivity by teachers and other school employees. Methods: A combination of agent-based and Monte Carlo economic simulation modeling was used to determine the cost-benefit of closing schools (vs. not closing schools) for different durations (range: 1 to 8 weeks) and symptomatic case incidence triggers (range: 1 to 30) for the state of Pennsylvania during the 2009 H1N1 epidemic. Different scenarios varied the basic reproductive rate (R 0) from 1.2, 1.6, to 2.0 and used case-hospitalization and case-fatality rates from the 2009 epidemic. Additional analyses determined the cost per influenza case averted of implementing school closure. Results: For all scenarios explored, closing schools resulted in substantially higher net costs than not closing schools.
5,7 for the HIVNET Vaccine Preparedness Study Protocol Team Recent studies have reported on the u... more 5,7 for the HIVNET Vaccine Preparedness Study Protocol Team Recent studies have reported on the utility of audio computer-assisted self-interviewing (ACASI) in surveys of human immunodeficiency virus (HIV) risk behaviors that involve a single assessment. This paper reports the results of a test of ACASI within a longitudinal study of HIV risk behavior and infection. Study participants (gay men (n = 1,974) and injection drug users (n = 903)) were randomly assigned to either ACASI or intervieweradministered assessment at their second follow-up visit 12 months after baseline. Significantly more of the sexually active gay men assessed via ACASI reported having sexual partners who were HIV antibody positive (odds ratio = 1.36, 95% confidence interval: 1.08, 1.72), and a higher proportion reported unprotected receptive anal intercourse. Among injection drug users (IDUs), our hypothesis was partially supported. Significantly more IDUs assessed via ACASI reported using a needle after another person without cleaning it (odds ratio = 2.40, 95% confidence interval: 1.34, 4.30). ACASI-assessed IDUs reported similar rates of needle sharing and needle exchange use but a lower frequency of injection. Participants reported few problems using ACASI, and it was well accepted among members of both risk groups. Sixty percent of the participants felt that the ACASI elicited more honest responses than did interviewer-administered questionnaires. Together, these data are consistent with prior research findings and suggest that ACASI can enhance the quality of behavioral assessment and provide an acceptable method for collecting self-reports of HIV risk behavior in longitudinal studies and clinical trials of prevention interventions.
To evaluate the effect of the assumption of no long reporting delays on estimates of AIDS inciden... more To evaluate the effect of the assumption of no long reporting delays on estimates of AIDS incidence. Reported AIDS cases must be adjusted for reporting delays to estimate AIDS incidence. We compared the adjustments supplied with the Centers for Disease Control and Prevention (CDC) AIDS Public Information Data Set with a set of adjustments that differ with respect to CDC assumption of no long delays. Both methods assume that no changes in delay have occurred throughout the reporting period. Probability distributions of reporting delays were calculated from the delay adjustment weights supplied by CDC, and from an alternative method that estimates the probability of long delays from the surveillance data. AIDS incidence estimates from these two approaches were calculated and compared. The CDC adjustments assume that there will be no reporting delays longer than 61 months, whereas the alternative method estimates that 5.9% of case reports will be delayed longer than 61 months. The CDC adjustments result in lower estimates of AIDS incidence and a flattening of the epidemic. In addition to a 6.2% reduction in total estimated AIDS incidence, the CDC assumption changes the shape of the estimated epidemic. These may result in as much as 4-21% reductions in model estimates of HIV incidence and prevalence.
Computational and Mathematical Organization Theory, 2015
Many published influenza models treat each simulation day as a weekday and do not distinguish wee... more Many published influenza models treat each simulation day as a weekday and do not distinguish weekend days. Consequently, the weekend effect on influenza transmission has not been fully explored. To assess whether distinguishing between weekday and weekend transmissions in simulation models of flu-like infectious disease models matters, this study uses an agent-based model of the Chicago, Illinois metropolitan area. Our study assesses whether including weekend effects is offset by increases in weekend contact patterns and if implementing 3-day weekends dampens disease transmission enough to warrant its use as a containment strategy. Results indicate that ignoring weekend behaviors without incorporating increases in community-based non-school contacts (i.e., compensatory behaviors) causes the peak case incidence day to occur 7 days earlier and can reduce the peak attack rate by as much as 60 %. These results are sensitive to the proportion of symptomatic cases that are assumed to remain at home until they recover. The 3-day weekend intervention has interesting possibilities, but the benefits may only be effective for mild epidemics. However, a 3-day weekend for schools would be less detrimental to the educational process than sustained permanent closing because student and teacher contact is maintained throughout the epidemic period. Also, a 4-day school and work week may be more easily accommodated by many types of schools and businesses.
Maternal nutrition influences a child’s birthweight, which affects the child’s growth and subsequ... more Maternal nutrition influences a child’s birthweight, which affects the child’s growth and subsequent survival. However, the broad consequences of maternal undernutrition and the outcomes of interventions to improve maternal nutrition take years to manifest. To examine the long-term health outcomes of low birthweight infants in response to a maternal nutritional supplementation intervention without this obstacle, we developed the Forecasting Population Progress (FPOP) microsimulation model. The intervention we assessed was based on the findings of a published clinical trial outcome that reduced the incidence of low birthweight, a known cause of stunting. We implemented the “before intervention” and “after intervention” simulations and generated the difference in outcomes, using a spatially explicit synthetic baseline population of Indonesia generated from a microdata sample of the Indonesian 2010 census. We focused specifically on two provinces—Yogyakarta and Bali—which represent dif...
systems for accurate and cost-effective prediction of human health outcomes. Development of incre... more systems for accurate and cost-effective prediction of human health outcomes. Development of increasingly sophisticated in vitro tissue models has accelerated in recent years, driven by the recognition that conventional two-dimensional (2D) cell culture formats do not adequately recapitulate the 3D arrangements of cells and extracellular matrix of tissues and 1 Introduction 1.1 Organotypic cell cultures An intense area of research aims at improving drug development and toxicology assessment by shifting away from traditional animal studies and towards human relevant, in vitro
Measurements of drug use and other illicit or stigmatized behaviors are subject to nontrivial und... more Measurements of drug use and other illicit or stigmatized behaviors are subject to nontrivial underreporting biases. During in-person surveys, respondents are more likely to report such behaviors when interviewed using techniques that maximize interviewee privacy, e.g., use of paper SAQs and audio-CASI rather than questioning by human interviewers. Until recently, respondents in telephone surveys could not be offered similar privacy. A new technology, telephone audio computer-assisted self-interviewing (T-ACASI) overcomes this limitation of telephone surveys by allowing respondents to respond to a computer. A randomized experimental test of T-ACASI was embedded in the Urban Men's Health Study (UMHS). UMHS surveyed a probability sample of 2,881 men from four United States cities and who reported having sex with men. Respondents interviewed using T-ACASI reported a higher prevalence of drug use and drug-related behaviors than respondents interviewed by human interviewers. However, survey respondents were more likely to break off an interview when the interview was conducted by a T-ACASI computer rather than by a human interviewer.
We present an infectious disease model with the following characteristics: • Eight disjoint risk ... more We present an infectious disease model with the following characteristics: • Eight disjoint risk groups that are based on transmission via sexual and intravenous drug use in homosexual men, bisexual men, heterosexual men and women. • Contacts that follow from a homogeneous mixing assumption which takes into account risk group sizes and behavior. • A disease incubation function based on the results of Bachetti and Moss (1989). • No infections associated with hemophiliacs, blood product users, children under 13 years of age, and foreign borns. Simulation experiments were conducted to estimate critical model parameters such as the number of sexual and/or needle sharing contacts needed to explain observed AIDS cases and deaths due to HlV-related illnesses. These estimated parameters were used to make short-term national projections of the incidence of HIV and AIDS. I. INTRODUCTION This paper presents a simple model of the AIDS epidemic developed to explore some fundamental relationships...
Uploads
Papers by Philip Cooley