Parkinson's Disease.

One disease, four entry points.

Parkinson's disease is not a single mechanism. The dominant clinical and biological model converges on four contributors: neuroinflammation centred on NF-κB signalling, mitochondrial dysfunction in dopaminergic neurons, gut-derived inflammation and α-synuclein seeding via the gut–brain axis, and impaired proteostasis with Lewy-body aggregation. Atlas Bio's pipeline assigns one blend to each axis and pre-specifies which patient phenotype each blend is expected to benefit.

Mechanism-anchored, phenotype-mapped.

Each blend's compound list, dose escalation, and combination logic are gated. Gated

Locked before readout.

The PD-NF-01 Phase IIa design carries a SHA-256-locked prediction set covering target engagement, cytokine shifts, motor scores (UPDRS-III), non-motor (sleep, autonomic), and progression-rate slope. The lock makes the data the final arbiter — if the predictions fail, the framework updates, not the predictions.

Hash, prediction wording, and timestamp are available to verified clinicians and PD researchers on request. Gated

The PD pipeline as a platform.

The four PD blends form the analytical backbone of three downstream programs:

• MS (RRMS) — MS-IMM-01 inherits the NF-κB modulation strategy; see MS abstract →
• ALS — repurposes PD-MITO-02 and PD-PROT-04 with ALS-specific dosing; see ALS abstract →
• MSA — alpha-synucleinopathy with autonomic primary endpoints; see MSA abstract →

Cross-walk modifiers (mechanism-anchored adjustments when re-fitting a blend across indications) are part of the gated methodology pack.

Same stack, every indication.

• M2 · Pre-Registration — SHA-256 locking before readout.
• M1 · IBC Reasoning — 19-agent audit (8 reasoning nodes + 11 core agents) of every prediction.
• M3 · Cross-Trial Validation — mechanism-anchored bridging from PD evidence base to derivative indications.
• M4 · RWE Pipelines — ClinicalTrials.gov, PubMed, FDA, EDGAR, and patient-reported feeds.