Parameter estimation for connectome generative models: Accuracy, reliability, and a fast parameter fitting method.

Area of Science:

• Computational neuroscience
• Network science
• Brain imaging analysis

Background:

• Generative models simulate human brain networks using probabilistic wiring rules.
• Parameter fitting quantifies geometry and topology's role in brain network formation.
• Current parameter estimation methods are computationally intensive and lack validation for large cohorts.

Purpose of the Study:

• To introduce a fast, reliable, and accurate parameter estimation method for connectome generative models.
• To address the computational burden of parameter estimation in large-scale connectome studies.
• To provide guidance on leveraging the accuracy-reliability-computational expense tradeoff.

Main Methods:

• Developed a novel, computationally efficient parameter estimation technique for generative connectome models.
• Validated the method's accuracy and reliability against established approaches.
• Empirically determined minimum sample sizes for detecting group differences in model parameters.

Main Results:

• The proposed method significantly reduces computational cost (by orders of magnitude) while improving estimation accuracy and reliability.
• An inherent tradeoff between estimation accuracy, reliability, and computational expense was identified and characterized.
• Empirical approximations for minimum sample size requirements were established for power analyses.

Conclusions:

• The new method offers a scalable and efficient solution for parameter estimation in connectome generative models.
• Understanding the accuracy-reliability-cost tradeoff is crucial for optimal application in large cohort studies.
• This work provides a practical statistical guide for applying generative models in neuroscience research.