Bayesian Networks with Latent Time Embedding for Stage-Aware Causal Modeling of Alzheimer's Disease Progression
Researchers have developed a new Bayesian structural framework called Bayesian Networks with Latent Time Embedding (BN-LTE) to model the progression of Alzheimer's disease. This model estimates disease pseudotime from baseline biomarker data and enforces biologically plausible ordering for the amyloid-tau-neurodegeneration (AT(N)) cascade. BN-LTE was evaluated using data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and demonstrated strong spatial reconstruction of tau progression compared to existing forecasting methods. The framework also identified a specific window during disease progression where amyloid sensitivity is most pronounced. AI