Three tools and a research library, one governance philosophy. Pick a tool to see what it does and how to engage with it.
AI Governor v1 evaluates clinical AI predictions against deterministic clinical scoring tools published in medical literature. KDIGO for AKI. Padua for VTE risk. C-SSRS for psychiatric assessment. APACHE II for ICU mortality. Each evaluation is cryptographically signed with Ed25519. Click through to see what each modality looks like.
Governance and audit infrastructure for clinical AI vendors. Not a clinical-decision-support device — not used on real patients. Inputs are synthetic, public-dataset, or de-identified.
Discordant chest x-ray case: AI device says "normal", human radiologist reads right lower lobe consolidation, early pneumonia. The pipeline runs imaging-modality probe → body-region specialist → 3-variant adversarial vision reasoning, returns a procurement-readable verdict with named verdict-provenance (primary · text_override · scope_gap_kept) and a methodology-audit grounding signal (grounded · partial · ungrounded · no_clinical_claim) listing which named criteria sets the reasoning referenced. ~45 s end-to-end.
Adam will show you the methodology, the verification flow, and what a sandbox evaluation looks like at your institution.
Start 30-day Trial View Pricing Contact AdamThe Discovery Engine is the research arm. It surfaces contradictions and patterns across published clinical mortality data. Two ways to engage: license the engine in your environment with monthly contradiction reports, or commission single-paper integrity scoring.
Cross-dataset replication for mortality, sepsis, and risk-stratification claims. Pre-registered methodology, OSF evidence chain.
Ed25519-signed evidence hashes for every finding. Public methodology. Confirmatory replication is pre-registered before clinical claims are made.
Dedicated engine instance, your data stays in your environment, monthly contradiction reports. Or single-paper Evidence Integrity Score on request.
Whether it is a license, a single-paper integrity scan, or a partnership conversation, Adam responds personally.
Start 30-day Trial View Pricing Contact AdamCaliper takes a drug-safety claim and runs it through a pre-registered validation pipeline against public adverse-event data. Disproportionality analysis with proportional reporting ratio, 95% confidence interval, and Yates χ2 — the same math regulators use. Inputs are hashed and locked the moment you submit, so the methodology cannot be moved after the data is in view. Every finding is Ed25519-signed and chain-anchored in an append-only ledger that anyone can verify in the browser via the public /keys/ page. Submissions that don't clear the gates publish as null result, not hidden. The framework publishes what fails just as carefully as what passes.
Inputs hashed and pre-registration sha1-locked at submission. The pipeline runs and publishes a signed verdict — positive or null.
Submit just a drug name. The engine enumerates the top 20 adverse events, validates each, and ranks the survivors by PRR.
Eleven curated classes — SGLT2 inhibitors, GLP-1 agonists, anti-amyloid mAbs, KRAS G12C, BTK inhibitors, S1P modulators, CGRP antagonists, and more. Fan out across every member drug.
Dedup ≥ 5 unique cases · suspect-drug filter · max co-exposure < 80% · PRR ≥ 2 with lower 95% CI > 1 · n ≥ 3.
Findings publish as candidate, then are promoted to audit-grade after operator review, or retired to null-result / withdrawn.
Each finding has a public verify page. Paste the identifier, see the evidence hash, prior-entry chain link, and Ed25519 signature. No login. No install.
Every finding ships as a downloadable tar.gz with the manifest, the canonical bundle, the rendered HTML fragment, and step-by-step verify instructions.
Findings that don't clear the gates are signed and published anyway — proof the platform doesn't bury what doesn't validate.
For real-world evidence teams, pharmacoepidemiology groups, signal-detection scientists, clinical-pharmacology directors, biotech R&D teams. Submit single drug+AE claims, run single-drug discovery scans, or run drug-class scans across curated cohorts. Every verdict cryptographically signed and browser-verifiable.
Start 30-day Trial View Pricing Contact AdamEvery paper here is approved by Adam personally. No auto-indexing. Each preprint is shared for pre-submission peer commentary and carries the full evidence chain — datasets, code, OSF preregistration, signed hashes.
Across 286,510 ICU encounters in MIMIC-IV and eICU-CRD, mainstream sepsis prediction benchmarks correlate with hospital billing patterns rather than biological sepsis criteria.
Read the full paper →Sepsis-AI tools trained on academic-medical-center data systematically misfire in community hospitals. The miscalibration is structural, not statistical, and pre-registered replication confirms it.
Read the full paper →Published mortality estimates for elderly ICU patients underestimate true risk by 66–168% across MI, sepsis, and respiratory-failure cohorts. Cross-validated on 201,905 encounters.
Read the full paper →The full library, including abstracts, OSF links, and download bundles, lives at /research.
Tell Adam which paper or which question, and he'll reply directly.