Patent Notice

This page is provided to satisfy virtual patent marking requirements under 35 U.S.C. §287(a) and equivalent provisions in other jurisdictions. The following products and features of the vasus.ai / exposomic.ai platform are the subject of one or more pending patent applications.

1. EHSPI™ — Environmental Health Sensitivity Performance Index

The EHSPI (Environmental Health Sensitivity Performance Index) scoring architecture — including the three-layer abstraction of (i) real-time environmental signal ingestion and condition-specific feature computation, (ii) evidence-weighted composite scoring across five chronic condition sensitivity dimensions, and (iii) confidence-graded relative location ranking — is patent pending.

2. Session-based privacy-by-design architecture

The session-based, privacy-by-design architecture for generating personalised environmental sensitivity real-time intelligence without storing any user data — using session-scoped sensitivity selections and anonymous session tokens only, with no persistent user identity — is patent pending.

3. Multi-stage biomedical retrieval and environmental health synthesis pipeline

The multi-stage biomedical retrieval and environmental health data context synthesis pipeline — combining contrastive-trained semantic encoding with asymmetric retrieval, a five-stage quality gate (query planner, anatomical prefilter, bi-encoder, cross-encoder, LLM gate), multi-factor evidence ranking, and constrained generative synthesis — all without storing user identity — is patent pending.

4. Systems and Methods for Provenance-Composed Environmental Health Intelligence with Deterministic-Integrity Agentic Architecture

Systems and methods are disclosed for a provenance-composed environmental health intelligence platform comprising seven inventive clusters of novel technical architecture. The platform implements: (A) a two-pass provenance-laddered composition rule that estimates personal intervention effects by combining population-derived priors with an n-of-1 personal event ledger, wherein numeric estimation is strictly upstream of any natural-language generation; (A/H) retrospective episode exposure reconstruction via one-directional wall-preserving spatiotemporal join of personal presence intervals against a permanently-stored place-state history, yielding per-episode exposure estimates with provenance-tiered confidence composable into the personal estimator; (B) exception-instantiated ephemeral investigator agents operating over deterministic multi-layer environmental surveillance, achieving order-of-magnitude compute reduction versus resident per-location agents; (C) governed generation of statutorily-grounded artefacts via deterministic context assembly, template-constrained language model drafting, and compliance verification with structural hallucination exclusion; (D) retraction-safe approximate vector retrieval implementing post-scan retraction exclusion that preserves partial index performance, avoiding the documented 94× latency regression caused by predicated vector-path queries; (E) anchored cohort-versioned environmental scoring with persisted normalisation parameters enabling cross-temporal comparability; and (G) a deterministic-ontology agentic facade architecture with consent-scoped subgraph sharing, wherein agentic access is exclusively mediated by schema-constrained tool facades and model writes are structurally prohibited. An overarching system claim (H) integrates population context, personal context, and hard separation between them under a shared provenance kernel.

Trademarks

vasus.ai™ and EHSPI™ are trademarks of the operating entity. Use of these marks without permission is prohibited. Trademark registration applications are pending in the UK and US.

Contact

For patent or trademark enquiries: ops@exposomic.ai

This notice is updated as patent applications are filed and granted. The list above may not be exhaustive. Other features of the vasus.ai platform may be the subject of additional pending or future patent applications. Last updated: 3 September 2026.