[Week 8–10] ADNI clinical validation: finite-horizon E* as disease stage biomarker

[TheRoy666]

ADNIPipeline is implemented and validated on synthetic cohorts. This issue
tracks deployment on real ADNI-3 data.

Tasks:

• Obtain ADNI-3 access (adni.loni.usc.edu) — rsfMRI + DTI cohort
• Run ADNIPipeline on CN / MCI / AD subjects (target: 30+ per stage)
• Generate stage_trajectory() for default_to_memory transition
• Run finite_vs_infinite_comparison() on representative CN and AD subjects
• Test: does finite-horizon E* show stronger stage separation than W∞?
• Compute DMN controllability decline across CN → MCI → AD

Core NeuroSim claim: finite-horizon metric detects AD-related impairment
that the infinite-horizon approximation misses.

Note: Retrogenesis analysis (the Entropic Mirror project) is separate and
out of scope for this issue.

Tracks GSoC Weeks 8–10 deliverable.