Purpose: Discuss a Modeling Plan, Then Its Results
A model-informed drug development (MIDD) approach can influence consequential choices: which dose to study, how to design a trial, or whether available evidence supports a development decision. Sponsors may benefit from FDA discussion before investing in a substantial analysis, but a proposal is not the same as results. The FDA MIDD Paired Meeting Program is designed to connect those two stages: discussion of a proposed approach and, later, discussion of the results and their implications.
The word paired describes the planned relationship between the meetings. The first is prospective: the sponsor explains a proposed MIDD application and seeks FDA feedback on the approach. After the sponsor carries out the work, a second meeting allows discussion of the results. The sequence can help focus the analysis on a defined development question and then return to that question with evidence in hand. It does not mean FDA has approved the method or agreed in advance with the result.
The program is relevant when the planned use of modeling is important enough to merit focused discussion and can be described concretely. The sponsor should be able to identify the decision the model is intended to inform, the evidence and assumptions that matter to that decision, and what feedback would help make the work more useful. The earlier material on context of use and decision fitness is useful here: a model’s technical sophistication alone does not establish that the proposed use is well framed.
Eligibility and Suitability
The program is for sponsors developing drug or biological products within CDER’s regulatory purview. Product jurisdiction matters: the program should not be assumed to apply to a product regulated by another FDA center. Sponsors should check the program’s current instructions and confirm the appropriate FDA review division and submission route before preparing a proposal.
Eligibility is only one part of fit. A candidate proposal should describe an identifiable MIDD application in a product’s development and a decision or uncertainty that the application could address. Examples may involve dose selection, exposure-response analysis, clinical trial simulation, or another modeling application with a plausible role in development or regulatory decision-making. The proposal should be sufficiently specific for FDA to discuss the planned use, rather than asking generally whether modeling is valuable.
A strong fit has several features:
- A defined decision: The sponsor can say what choice the analysis is meant to inform and when that choice will arise.
- A discussable plan: The proposed data, model use, assumptions, and intended interpretation can be summarized before the analysis is complete.
- A reason for paired discussion: Feedback before the work could influence the plan, and the results will later raise a meaningful question for discussion.
- A feasible sequence: The sponsor can reasonably conduct the analysis and return with interpretable results within a development timeline.
These features are practical indicators of suitability, not a guarantee of selection. A proposal may be scientifically worthwhile but poorly suited if it has no defined decision, if the analysis is already complete and only retrospective feedback is sought, or if it asks FDA to endorse an unspecified modeling strategy. Conversely, an analysis need not be novel in every technical detail to be useful; its value depends on the proposed application and the decision it may inform.
Worked Example: Is a Dose-Selection Proposal a Good Fit?
A hypothetical sponsor developing a CDER-regulated medicine plans to use exposure-response modeling to compare two dose regimens before finalizing a confirmatory trial. The analysis will combine existing trial data with a proposed simulation. The sponsor can identify the dose-related decision, describe the planned analysis, and state which assumptions—such as the exposure-response relationship and the relevance of the patient population—could change that decision.
Eligibility check. The hypothetical product is stated to be under CDER’s purview, so it meets the program’s center-level scope. The sponsor must still follow the current program instructions and applicable meeting procedures; this fact alone does not establish selection.
Suitability check. The proposed modeling has a defined development use: informing dose-regimen selection for a confirmatory trial. The first meeting could address whether the planned application and its assumptions are framed clearly enough for the intended decision. A later meeting could examine the results and whether they support the sponsor’s planned interpretation.
Conclusion. This scenario appears well suited for consideration because there is a prospective plan and a consequential decision to discuss both before and after the analysis. FDA feedback would inform the sponsor’s work; it would not itself select the dose or guarantee acceptance of the eventual analysis.
The Paired Sequence and Its Outputs
The program’s central sequence has two substantive discussions. Program-specific submission and scheduling steps should be checked in current FDA instructions; the stages below describe the logic of the paired interaction, not a promise that a proposal will be accepted or that a particular date will be available.
Describe the product and proposed MIDD application, the development decision it may inform, the planned analysis, and the questions for FDA. Follow the program’s current submission instructions.
In the first meeting, present the rationale and plan. The discussion can surface concerns about the intended use, data, assumptions, or interpretation before the sponsor completes the work.
The sponsor performs the analysis and records the data, methods, assumptions, diagnostics, limitations, and results needed to interpret it in context.
The later meeting focuses on what the analysis found, how robust and relevant the findings are to the stated decision, and what questions remain. The timing and materials should follow the applicable FDA process.
For the first discussion to be useful, the sponsor should make the proposal reviewable. A concise package can state the decision and context of use, summarize the data and modeling strategy, identify key assumptions and uncertainties, and explain what alternative results would mean for the decision. Questions should invite actionable scientific feedback—for example, whether a proposed population or endpoint is appropriate for the intended inference—rather than simply request general approval.
For the second discussion, results should be presented with enough context to assess their meaning. That includes relevant diagnostics and limitations, not only a preferred estimate or simulation. If the results differ from expectations, the sponsor should explain which assumptions, data features, or model limitations may account for the difference. The meeting is an opportunity to discuss the evidence and its proposed use, not a mechanism for converting an uncertain result into a binding determination.
The meeting record, including FDA’s meeting minutes under the applicable process, helps document the discussion and feedback. Sponsors should distinguish that record from a formal approval, a binding decision, or an assurance that a future submission will be accepted. Feedback is specific to the questions and information presented; conclusions may change if the data, methods, product context, or proposed use later change.
Reasoning Through Program Fit
A practical fit assessment can be derived from the program’s two-stage structure. If the first meeting is to improve a proposed analysis, there must be a plan that can still be discussed and potentially refined. If the second is to interpret results, there must be a feasible path to produce results that address the same decision. Thus, the sponsor should test both halves of the proposal rather than asking only whether the modeling topic sounds relevant.
This is a reasoning aid, not a numerical eligibility rule. Each component is necessary for a coherent paired rationale: without a decision, the model’s relevance is unclear; without a plan, prospective feedback has no defined target; without interpretable results, the second discussion cannot close the loop. Even when all three are present, FDA selection remains a separate matter.
Worked Example: A Proposal With No Clear Decision
A hypothetical sponsor proposes a complex disease-progression model for a CDER-regulated product. The sponsor asks FDA to discuss the model’s general validity but does not identify a planned dose, trial-design, labeling, or other development decision that the results could inform.
Apply the fit test. The product’s stated jurisdiction is consistent with the program’s scope. However, the proposal does not specify what choice depends on the model. Therefore, FDA cannot readily assess which model outputs matter, what degree of uncertainty is consequential, or what result would change the sponsor’s action.
Improve the proposal. The sponsor could define a decision, such as whether a planned trial should use one of two specified follow-up durations, and explain how model predictions would inform that choice. It would then need to specify the relevant data, assumptions, and intended interpretation. If the model is not expected to inform a concrete development decision, another interaction or a more focused proposal may be more appropriate.
Conclusion. As originally described, the proposal is a weak fit—not because disease-progression modeling is inherently unsuitable, but because the proposed use is not decision-linked.
How It Differs From Other FDA Interactions
A standard FDA formal meeting can address a range of development questions under the applicable meeting framework. The MIDD Paired Meeting Program is distinctive in its structured link between a proposed modeling approach and later discussion of the results. It is not simply a new name for every meeting that includes a model, and it does not displace the ordinary processes for advice, submissions, or regulatory decisions. The earlier material on FDA formal meeting types provides the broader context for choosing a formal interaction.
| Feature | MIDD Paired Meeting Program | Other formal FDA meeting |
|---|---|---|
| Organizing purpose | Discuss a proposed MIDD application and later its results as a linked sequence | Address the specific questions appropriate to that meeting and its stage |
| Best rationale | A decision-linked modeling plan that benefits from prospective and results-focused discussion | A defined regulatory or development question that does not require the program’s paired structure |
| What it does not provide | Automatic model approval, guaranteed selection, or a binding decision | Those assurances are not inherent in a meeting; the status of feedback depends on the applicable regulatory process |
A sponsor should not choose the paired program merely because a model appears in the meeting package. If the immediate need is a discrete question and no later results discussion is planned, a different formal interaction may be more proportionate. If the sponsor needs both early discussion and a planned return after analysis, the paired structure may be advantageous. The choice depends on the question, timing, and expected outputs, not on a claim that one route is universally superior.
Failure Modes and Limitations
- Confusing participation with endorsement. Being selected for the program, or receiving feedback in a meeting, does not mean FDA has endorsed the model or agreed with a future regulatory conclusion.
- Submitting a broad topic instead of a question. “We want to discuss modeling” gives little basis for focused feedback. State the decision, the proposed use, and the uncertainties that could affect it.
- Planning only for the first meeting. If no feasible analysis or second discussion is anticipated, the paired format may add little. Explain how the work will produce interpretable results relevant to the same decision.
- Presenting results without limitations. A preferred model output without diagnostics, assumptions, and applicability limits may not support a useful discussion. Explain what the analysis can and cannot establish.
- Assuming the program overrides other processes. Program participation does not remove applicable submission requirements or make a meeting record a binding decision. Use the appropriate regulatory pathway for the specific question.
The program also has limits: selection is not assured, meeting time is finite, and advice depends on the information supplied. A model may be too early, too poorly specified, or too disconnected from a development decision for the proposed interaction to be productive. The sponsor remains responsible for the analysis and for deciding how to respond to feedback; FDA’s discussion does not transfer that responsibility.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- What makes the program’s meetings “paired”?
- A sponsor has a model but cannot name a development decision its results would inform. What is the main fit problem?
- Does feedback in the first meeting establish that FDA will accept the final model-based conclusion? Why or why not?
- A sponsor has completed an analysis and wants only retrospective feedback, with no proposed approach left to discuss. What feature of the paired program is missing?
- Name two items that should be explained when presenting results at the later meeting.
Solutions. (1) The first meeting discusses the proposed MIDD approach; a later meeting discusses results after the work is conducted. (2) The proposal is not decision-linked, so it is unclear which outputs matter or how feedback would improve the development plan. (3) No. Meeting feedback is not automatic endorsement or a guarantee that a future submission or conclusion will be accepted. (4) The prospective discussion of a proposed approach is missing, so the two-stage rationale is absent. (5) Examples include the data and methods, key assumptions, diagnostics, limitations, and how the findings relate to the stated decision; any two of these, explained in context, are appropriate.