Standard causal-inference tools report a point estimate. Rosenbound reports the estimate, the Rosenbaum-Γ sensitivity bound that says how fragile it is, a five-method robustness pentagon, and a 21 CFR Part 11 ALCOA+ audit chain — on one platform, built for the pharmacovigilance and real-world-evidence teams who answer to the FDA.
“Identifying and addressing the presence of confounding and other forms of bias is critical… planned sensitivity analyses to assess the robustness of study findings.”
— FDA, Non-Interventional Studies draft guidance, March 2024
The FDA Sentinel Innovation Center’s PRINCIPLED process (Desai, Wang et al., BMJ 2024;384:e076460) prescribes “a plan for robustness assessments including deterministic sensitivity analyses, quantitative bias analyses, and net bias evaluation.”
Rosenbound’s Rosenbaum-Γ bound is precisely this class of robustness instrument — reported as a default output on every study, not an afterthought. It is the operational answer to a stated regulator expectation.
Aligned with the FDA RWE Framework (final guidance Aug 31, 2023) and the EHR/Medical Claims Data guidance (final Jul 25, 2024). Full alignment map at /regulatory.
Watch the platform walkthrough at rosenbound.ai — three moments: the Cognitive Validation Report refusing incoherent data, the live Γ-bound sensitivity visualization, and the reproducibility certificate generated on every study. The full platform is open to Design Partners.
Watch the preview →The same causal-inference engine and audit ledger, across four settings. Pharmacovigilance and real-world evidence are live; drug discovery and physician decision support are on the roadmap. Each vertical has its own page.
For drug-safety and pharmacovigilance teams in pharma and at CROs. Replaces hand-coded FAERS-disproportionality workflows. Every adverse-event signal arrives with Rosenbaum bounds, MedDRA preferred terms, a dual-attestation reviewer workflow, and 21 CFR Part 11 ALCOA+ provenance — regulator-ready by default.
For RWE methodology leads at academic medical centers and CROs. Five-method causal sensitivity pentagon as a single API call. Cohort DSL, OMOP CDM ingestion roadmap, target-trial-emulation protocol scaffolding, reproducibility certificate on every study. Replaces the hand-rolled methodologist work behind every observational study.
For computational biology and target-discovery teams. The D-MPNN molecular-property stack (validated on BACE / BBBP / HIV scaffold splits) is benchmark-locked. Causal mechanism-of-action work is on the 2027 roadmap; the benchmark capability stands as a published claim today.
For ICU and academic-medical-center physicians making individualized treatment decisions. Targeted FDA Class II SaMD via 510(k) with a Predetermined Change Control Plan; Q-Submission planned after the first Design Partnership concludes. Available post-2027 after clinical-deployment validation.
Behind the login — and inside the live demo at rosenbound.ai — is a working clinical-AI platform. Five modules, integrated, mutually auditable. This is the platform a Design Partner uses on day one.
CSV upload with schema auto-detection, type inference, archetype suggestion, treatment/outcome/index-date mapping into an auto-generated Cohort DSL. The Cognitive Validation Report blocks ingestion of incoherent data and explains why — temporal, ontological, biological, and causal-acyclicity vetoes.
Submit a cohort, get five causal estimators on the same data: AIPW, doubly-robust ATT with overlap trimming, IV-LATE, neural counterfactual, and Rosenbaum bounds. Interactive Γ-slider on every run with the crossing-point explicit. Per-method drill-down with covariate balance + feature attribution.
Adverse-event signals with MedDRA preferred terms, SOC class, severity score, Γ-bound per signal, and a reviewer state machine (queued → triaged → in-review → quality-review → escalated/dismissed → signed-off). Filter by SOC, drug, term, severity, reviewer. Dual-attestation on high-severity sign-offs.
Every state transition cryptographically committed to a SHA-256 chained ledger. ALCOA+ aligned (Attributable, Legible, Contemporaneous, Original, Accurate, Complete, Consistent, Enduring, Available). External-verifiable. Read-only export to pharma audit teams. The ledger is built to sit inside a regulated workflow without bolt-on audit infrastructure.
Per-tenant dashboard for active studies, signal queue health, and reviewer throughput. Every completed study generates a reproducibility certificate: cohort definition hash, cohort data hash, certificate ID, git commit, pinned library versions — the exact re-derivation guarantee regulators ask for, in the UI.
Standard practice reports one ATT estimate. Rosenbound reports the full sensitivity pentagon plus quantitative bounds — every number arrives with the uncertainty its method admits. Overview below; the full specification is at /methodology.
Doubly-robust per Robins-Rotnitzky-Zhao 1994 + Bang-Robins 2005. Three covariate-enrichment stages.
Hahn-1998 doubly-robust ATT estimand. Tight propensity clipping for small-N panels.
Per-prescriber preference instrument. Strong-instrument diagnostics + m-of-n bootstrap CI.
Patent-pending individual-level treatment-effect estimation with representation-balanced learning. Architecture under NDA.
Quantifies the unobserved-bias strength required to flip an estimate. The differentiating reporting layer.
Every figure below is from a full-corpus training run — not a smoke run, not a chosen seed. Headline lanes here; the complete tables (W1–W13, FAERS, ACIC22, D-MPNN, HNSI) are at /benchmarks, with raw log references available to Design Partners under NDA.
Pharmacovigilance triage satisfies all four criteria of 21st Century Cures Act § 3060: transparent algorithms, documented provenance, human-in-the-loop, independent review of the basis for the recommendation. Available to Design Partners in 2026 without 510(k).
The audit module is built to 21 CFR Part 11 § 11.10(a)–(k): SHA-256 chained ledger, monotonic timestamps, fsync + multi-writer protection, external-verifiable. Its outputs are structured to drop into a regulated workflow without bolt-on audit infrastructure.
Structured around the January 2025 draft “Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making” — context-of-use definition, model-risk assessment, credibility-assessment plan, execution, documentation, adequacy determination. Applied today on the PV and RWE verticals, not deferred to a future SaMD.
Q-Submission scoping for clinician-facing CDS is planned after the first Design Partnership concludes. A Predetermined Change Control Plan is drafted in parallel — the no-LLM-in-the-prediction-path, frozen-feature architecture keeps the modification space bounded and auditable.
Every state transition is committed to a SHA-256 chained ledger. External-verifiable. Tamper with any payload below — the hashes recompute live in your browser.
| # | Event | Payload | SHA-256 | Edit |
|---|---|---|---|---|
| 01 | cohort.create | cohort=illustrative-A · source=synthetic · dsl=validated | 0a5b506757…3290 | |
| 02 | study.run | study=illustrative-A · methods=AIPW, DR-ATT, IV-LATE, neural, Γ | d8e7e50348…ed6a | |
| 03 | signal.triage | signal=illustrative · meddra_pt=assigned · severity=high | a466c54b8d…a8b8 | |
| 04 | signal.signoff | signal=illustrative · dual_attest=true · state=signed_off | 35a2e96c2a…344e | |
| 05 | certificate.issue | certificate=illustrative-A · git=pinned · libraries=pinned | 1b871b2ad4…d338 |
pip install rosenboundProgrammatic access to the audited platform — for CRO methodology leads, drug-safety directors, and RWE teams who script cohort uploads, run sensitivity-bounded studies, and pull reproducibility certificates without leaving their analysis environment.
py.typed
View on PyPI →
get_current_user middleware. Tokens are JWTs minted from the platform’s API-key management surface (admin role required). Tokens are never in the SDK source or in PyPI metadata.require_permission(...) dependency on every protected route. Endpoints declare their required permission inline; missing permissions return HTTP 403 with no data echo.TenantContextMiddleware + Postgres Row-Level Security policies. Cross-tenant reads and writes are server-rejected even with a valid token. The platform reads nothing outside the caller’s organization.AppendOnlyAuditLog entries + a mirrored audit_logs row + 21 CFR Part 11 ALCOA+ attestation fields. Ledger writes are fail-loud: a write that cannot anchor to the SHA chain rolls back the entire request.cert = client.certificates.get(study_id) pdf = client.certificates.download_pdf(study_id)Hand the methodology PDF and the SHA-256 reproducibility hash to your QA team or an inspector. No screenshots, no email forwarding.
client.cohorts.create(csv=..., dsl=...)Iterate a directory of CSVs, call it for each, and capture the validation status. Fail-loud on any rejection so your pipeline halts instead of silently dropping rows.
client.studies.create(...).run().get_results()Wired into your existing Airflow / Prefect / cron stack. Pull pentagon estimates programmatically; route the Γ-bound + sensitivity intervals into your downstream dashboards.
Getting an API token. Design Partners receive scoped API tokens during onboarding — issued from the platform’s admin surface, tenant-locked, limited to the RBAC roles each team member needs. → Become a Design Partner
Five inventive concepts covering causal inference, audit-trail integrity, capability-aware abstention, modular cognitive substrate, and hybrid neuro-symbolic clinical evidence extraction.
— USPTO provisional filing
Harsh Singh filed a USPTO provisional patent on March 22, 2026, sole inventor, covering the methodology stack behind Rosenbound. The twelve-month window to non-provisional or PCT runs through March 22, 2027. Clinical AI is the first application of the underlying architecture. Algorithmic and architectural detail is available to Design Partners under NDA. Capability-level description is shared openly; implementation specifics are held back.
Rosenbound is selecting a small number of design partners across pharma drug-safety teams, CRO methodology groups, and academic RWE units. A Design Partnership costs nothing and moves no data: the platform provides production-shape synthetic cohorts; your team runs the full workflow — intake, pentagon, Γ-bounds, sign-off, certificate — in the live environment and tells us where it falls short of what a regulator would accept.