Connectome-based biophysical models of pathological protein spreading in neurodegenerative diseases
Fig 2
The unified framework for model evaluation.
Database: The unified database includes both raw data and parameter libraries, which are divided into individual-level data and population-level data, animal and human brain images, cross-sectional data and longitudinal data with multiple time points. Models: These data are used for model f(x), where researchers select appropriate biophysical models (NDM, ESM, and SIR) for individual-level or population-level analysis as appropriate. The selected model was fitted using the data provided by the parameter library and the simulated propagation process of pathological proteins/brain atrophy was obtained. Evaluation: The simulation results are compared with the real propagation process of pathological proteins/brain atrophy from both temporal and spatial perspective, utilizing measures of model performance (e.g., Pearson’s r, MAE), significance (spatial autocorrelation-preserving null models), and importance of model parameters (null models). The human and mouse icons are free icons provided by Microsoft PowerPoint.