Item 1: Title and Abstract – Identify in the title or abstract that the analysis depends on an individual-based mathematical model.

Whether the structure of a mathematical model can be identified depends on how it is indexed in the literature, which relies on an informative title [12] and abstract. By indicating in the title or abstract that the analysis involves an individual-based mathematical model, authors alert readers to be aware of particular assumptions, structure, and details in the body of the paper.

Item 2: Objective – State the objectives of the analysis with specific reference to the population(s), intervention(s), and time period(s) of interest.

Objectives should address the questions that the mathematical modeling exercise aims to answer, and reflect the efficacy, feasibility, and/or affordability of a particular HIV treatment or prevention intervention. A specific and focused purpose helps frame the paper’s methods and clarify the paper’s goals. The specificity of the objectives, especially with respect to the population, intervention, and time period of interest, justifies the model structure and indicates the data needs of the analysis. The information on setting and population will be essential for readers to assess the applicability and generalizability of the mathematical modeling results [12].

Item 3: Context – Justify the exploration of the policy question in the context of previous trials, cohorts, and modeling analyses for the time period(s) of interest.

Authors should explain how their analysis adds to the existing literature by noting previous trials, studies, and modeling exercises that address similar questions. The political and social context of the tested intervention should also be discussed, to inform the readers of the potential consequences of this exploration [5]. The need for a new model should be justified if there is expansion on an existing model or model structure.

Item 4: Model Justification – Explain the need for an individual-based mathematical model in the context of the objectives by referencing necessary model features (e.g. a need for heterogeneity of individual behavior and/or biology, and/or the explicit modeling of the interaction between individuals)

The reasoning for the use of an individual-based model should be clarified early in the paper structure. Because our reporting guidelines are specific to mathematical analyses with a specific intervention, the need to justify a model’s design becomes necessary. Authors should emphasize the need for an individual-based model design by discussing the necessity of heterogeneity as well as explicit interaction of individuals in the context of the objectives, or other model necessary model features they have incorporated that are essential to their analysis [8,10,13]. .

Item 5: Structure – Describe the model’s structure in both words and figures and describe how it affords the ability to explore the question(s) of interest.

By indicating the model structure in both words and figures, the authors are able to communicate effectively with the readers about the capability of the model structure and the validity of the model’s assumptions [8]. Figures are important for those unfamiliar with mathematical modeling, as they give a visual representation of what is happening inside of the model structure. Avoiding the “black-box” phenomenon will allow readers to better judge the quality of mathematical models in the literature and the subsequent results that arise from them [14]. Without a clear understanding of the structure of a given mathematical model, readers will have a difficult time piecing together which analyses and explorations are feasible.

Item 6: Assumptions – State the assumptions implicit in the model structure and justify with knowledge and data from the population of interest.

The generalizability of the conclusions drawn from the model analysis are dictated by the model’s assumptions [13]. Data used to justify assumptions should be included, so readers are aware of all limitations of the modeling approach. Simplifying assumptions related to the interaction of individuals in the population and the progression of HIV should be highlighted, as these have a direct impact on the authors’ ability to make accurate inferences.

Item 7: Validity of Sexual Behavior – Justify the validity of the necessary behavior accounted for in the model.

Many aspects of sexual behavior are important to the transmission and prevention of HIV in a population, and the level of detail inherent in individual-based models with respect to sexual behavior is much higher than any other type of modeling [15]. Authors should discuss all relevant behavioral processes including, but not limited to: relationship types, relationship durations, directionality in men who have sex with men (MSM) partnerships, number of sex acts per time step, presence or absence of risk groups, mixing pattern, mechanism of acquisition of partnerships, age of sexual debut, change in sexual behavior with aging or time, presence or absence of migration, and the presence or absence of testing and the treatment cascade. The elements of sexual behavior included in the model will depend on the analysis and population of interest and should be justified using data and information from the population of interest [8]. Authors should acknowledge poorly understood behaviors and limitations in data used to inform parameters.

Item 8: Validity of Biology - Justify the validity of the necessary biology accounted for in the model.

The level of biological detail that an individual-based model should represent is dependent upon the aims of the analysis of interest. Many elements of individual- and community-level biology are important to the transmission and prevention of HIV, for example: the inclusion or exclusion of sexually transmitted infection, the tracking of virological markers (CD4/HVL), multiple HIV disease stages with differing transmission probabilities, impact of circumcision status on HIV transmission, presence or absence of opportunistic infections, the impact of treatment on health and future transmission events, and the presence or absence of resistance mutations. By describing the elements incorporated in the model, the readers are able to determine which aspects of HIV transmission and prevention can be assessed.

Item 9: Parameters - List fixed parameters and calibrated parameters with ranges justified by the literature.

When individual-based models begin to represent sexual behavior and biology, the number of parameters needed to populate them grows dramatically [10]. By listing the main parameters examined in the analyses of interest, the readers are able to understand the necessary sources of the data. Further, including uncertainty ranges establishes the need for attaining parameter values through a fitting or calibration procedure (see Item 13) when data is not available from the literature. Where parameter values cannot be based on the literature or are not calibrated, data from similar populations or assumptions made about the population of interest should be provided [7,8].

Item 10: Time Step - State and justify the length of the time step used to advance model dynamics, if applicable.

The time step in an individual-based model is determined by the level of detail desired and computational limitations [13]. A short time step will provide greater detail on sexual behavior and biology, but will cause longer run times. Explicitly stating the time step, where appropriate, will make clear which processes the model describes well and makes the necessary data conversions for parameters transparent (see Item 19).

Item 11: Heterogeneity - Discuss how biological and behavioral heterogeneity is implemented in the model structure and whether this implementation allows for flexibility and specificity.

One of the main strengths of the individual-based model structure is the ability to represent heterogeneity in behavior and biology [3] and, as such, methods used to implement this heterogeneity need to be detailed. Specific details on the discrete categories or continuous distributions used should be reported in the manuscript or supplementary material. Emphasizing the particular elements of behavior and biology that vary across individuals allows readers to understand how accurately these processes reflect reality and highlights strengths of the modeling exercise.

Item 12: Interaction - Describe the parameters used to implement individual interaction in the model and justify the data used to parameterize these parameters.

HIV transmission that occurs within a particular partnership needs to be modeled accurately. Partnership formation and dissolution, the number of sexual acts in each partnership, the directionality of MSM partnerships, and the types of partnerships should be described in detail. By putting emphasis on the interaction between individuals in an individual-based model, authors will clearly describe the level of interaction between agents and consequently justify the need for this interaction.

Item 13: Calibration - If the objective of the analysis is to describe or predict dynamics in a particular population of interest, describe the process used to calibrate the model dynamics to existing data including the statistical procedure, the types of outcome measures used, and the quality of the data used

The process through which the model’s predictions, with regard to particular outcomes, are matched to data in the population of interest is called calibration [9]. There are many methods through which calibration can be statistically rigorous [5,9] and other methods through which it can be performed less rigorously. Calibration may be the most important piece of reported information for inferring a model’s ability to make accurate inferences about a particular population; therefore, details on the algorithms used and the outcomes calibrated should be provided. Although a model only needs to be calibrated once to reflect dynamics in a particular population, the calibration process should be repeated if the population, time period, or outcome of interest change. Models that have previously been calibrated should cite relevant previous manuscripts, and briefly describe the process. Authors should also make note of any effort to avoid over-fitting the model to data [16].

Item 14: Sensitivity Analyses - Summarize the results of sensitivity analyses on the main model parameters, discuss whether the results support the robustness of findings, and describe future work needed

As with any statistical analysis, it is essential to understand the sensitivity of the results to perturbations in parameters that are directly related to the intervention of interest [7]. For example, a model investigating how vaccine coverage impacts HIV incidence should vary vaccine coverage and efficacy to understand how model structure and parameterization impact the results. Sensitivity analyses may identify potential areas of model improvement and these discoveries should be noted. Less important results can be reported in supplemental material or an appendix.

Item 15: Assumption Sensitivity - Discuss how the behavior and inference of the model changes when particular assumptions (e.g. alternative mixing patterns, different levels of heterogeneity for behavior and biology) are altered or deleted

After describing the assumptions inherent in the model structure (see Item 6), the authors should discuss how the model’s behavior was dependent on these assumptions [8]. Assumption sensitivity analyses will reinforce the necessity of individual-based model structure and highlight which aspects of the intervention other model types would not capture. This type of sensitivity analysis will allow the authors to explore the generalizability of the model results to situations where the assumptions are violated [7,9]. Modelers will usually alter the assumptions most relevant for the analysis being performed, but altering other assumptions will improve the plausibility of the model results. Assumption sensitivity analyses can be performed in many different ways, one of which compares the results from the individual-based model to a deterministic model with simplified dynamics.

Item 16: Stochastic Sensitivity - Summarize the impact of stochasticity on the model runs and justify through random seed variation and sample size variation.

Microsimulation models are often stochastic, meaning there is a certain level of randomness inherent to each model run [9,13]. Stochasticity can affect large simulation models in different ways, depending on which processes rely on random number generation. Authors should describe how stochasticity affects model results and what the authors have done to understand these effects, including increasing sample size or seeding the population differently [7].

Item 17: Internal Validity - Describe the validity of the model programming by discussing how model bugs and program issues were checked and if modifications to model implementation were explored.

The internal validity of the model should be discussed, highlighting the steps taken to debug the model program and check the validity of the model structure [7,10]. This information should be placed in an appendix or supplemental material. Publications on previously published models should refer back to the methods of the original manuscript, but should not necessarily repeat the internal validity checks, unless the program code was changed. It is good practice to have two individuals program the model independently and compare the implementation to avoid bugs [7], though we acknowledge limited personnel or resources may make this infeasible. Model flowcharts, debugging runs, and other measures can also prevent bugs in the implementation of the code.

Item 18: Data Quality - Describe the quantity and quality of the data used to inform parameters for the population(s) of interest.

As mentioned in Item 9, the number of parameters needed to populate an individual-based model is large, and it may be difficult to find data for the parameters needed. A discussion of the data quality for the population and region of interest will allow readers to understand the limitations to modeling in this population [10] and may encourage public health researchers and social scientists to collect additional relevant data.

Item 19: Data Conversion - Discuss issues related to the conversion of data to fit the time step used.

The time step used is not often motivated by the data available, but rather by the goal of realism set by the authors. Data often has to be converted to the proper time step (e.g. number of sexual acts per partnership per time step). Any additional assumptions required for conversion should be discussed.

Item 20: Results - Present key modeling results with uncertainty estimates and indicate how many parameter sets were run for each analysis.

Authors should report their results with uncertainty estimates. Particularly for individual-based models, authors should disclose the number of parameter sets used or runs averaged to get the results. Discussing uncertainty in results will help the authors anchor their conclusions (see Item 23) and give the readers a better understanding of the capabilities of the model.

Item 21: Limitations and Strengths - Provide the key limitations and strengths of the modeling study.

The strengths and limitations of the methods used should be highlighted [12]. The limitations of an individual-based model are often dependent on the assumptions made, the computational power available, and the data used to inform the parameters. Detailed models of sexual behavior and transmission should emphasize the strengths of these details, while recognizing the potential weaknesses in data used to inform these processes.

Item 22: Reproducibility - Discuss whether the model is able to reproduce the behavior of other populations or interventions of interest.

The generalizability of results (discussed in Item 23) should be emphasized along with the generalizability of the model structure. Some models are flexible enough to describe the behavior of many populations, while others are best suited to the dynamics in a single population. Describing whether the software and programming needed for implementing the model structure can be used to answer other questions of interest should be noted. A description of the generalizability of the model may encourage collaboration with other modeling groups or authors who have interest in using the model structure to answer alternative questions.

Item 23: Discussion – Interpret the modeling analysis within realistic bounds, with reference to previous modeling studies, a discussion about the generalizability of the modeling results, and implications for future studies or models

The discussion section of a modeling paper should emphasize the capability of the model to represent real world dynamics, while keeping the conclusions grounded upon the model assumptions. The generalizability of the results should be discussed, with a particular focus on the assumptions that allow for generalizability of the findings [12]. Future modeling and non-modeling studies should be proposed with insight as to how this body of work would contribute to the HIV literature as a whole.

Item 24: Authorship and Funding - List sources of funding and describe each author’s contribution to the modeling framework and conceptualization.

Listing of funding sources allows other modelers to better understand what types of funding sources are applicable to modeling projects and grants. Additionally, by listing all the authors’ contributions, additional modeling teams or task forces can be composed based on capabilities of authors on previous projects and analyses.