Assessing the Role of Model Complexity in Virtual Clinical Trial Outcomes

Area of Science:

• Computational biology
• Pharmacometrics
• Mathematical modeling in drug development

Background:

• Virtual clinical trials (VCTs) offer potential for drug development but their reliability hinges on poorly understood design choices.
• Model complexity, prior parameter distributions, and virtual patient inclusion criteria significantly impact VCT outcomes.

Purpose of the Study:

• To investigate the influence of mathematical model complexity on VCT outcomes.
• To assess the impact of different prior parameter distributions and virtual patient inclusion methods on VCT results.

Main Methods:

• A case study of oncolytic virotherapy for murine tumors was used.
• Three mathematical models of increasing complexity were compared.
• Uniform and normal distributions were used as prior parameter distributions, with accept-or-reject and accept-or-perturb inclusion methods.

Main Results:

• The simplest model inadequately captured inter-patient heterogeneity.
• Intermediate and complex models showed diminishing returns in capturing patient response ranges.
• The accept-or-perturb inclusion method yielded more robust results, less sensitive to prior assumptions, compared to accept-or-reject.

Conclusions:

• VCT design should balance biological detail with model complexity, avoiding unnecessary intricacy.
• Employing inclusion criteria that avoid over-constraining patient populations is crucial for robust VCT predictions.
• The accept-or-perturb method is recommended for more reliable VCT outcomes.