Skip to content
Skip to content

NextGenAMR — a product by Vanguard Biotech Systems · Product assurance

Security and deployment

This page summarises the current approach to security, access, traceability and deployment for NextGenAMR. Some capabilities depend on the environment and the configuration agreed for a given deployment. Security controls are not a claim of clinical validation or regulatory conformity.

Product status:Being validatedContent version: v1.0.0Last updated: 2026-07-20

NextGenAMR is a functional platform in controlled preview. Validation is ongoing, and the product is not presented as an autonomous diagnostic or treatment-decision system.

Current controls

Only controls whose evidence supports publication are shown, each with its status and where it applies. States are shown as text, never colour alone.

Identity & access

  • Authentication & multi-factorImplemented

    Credential authentication with a password-strength policy and hashed passwords. Time-based one-time-password (TOTP) multi-factor with single-use recovery codes, per-account lockout after repeated failures, and server-enforced step-up re-authentication for sensitive actions.

    Scope: Multi-factor is enforced for privileged (administrative) roles and can be required per institution; it is available and enrollable for all users, but is not globally forced on every account.

    Limitations
    • · Multi-factor is not forced on every non-privileged account by default.
  • Session management & revocationImplemented

    Short-lived access tokens with rotating refresh tokens and server-side revocation: an administrative action (suspension, device revocation, institution block, logout-all) invalidates live sessions on the next request without waiting for expiry. Refresh-token reuse is detected and revokes the token family.

    Scope: Applies to all authenticated sessions.

  • Device registration & approvalImplemented

    New devices require explicit administrative approval before a session is issued, and can be revoked at any time (revocation ends sessions and is audited). Devices can prove possession of a private key at login, with optional mutual-TLS; binding strength is graded honestly, and mobile companions receive a scoped, read-only session.

    Scope: Cryptographic device-proof at login is supported and configurable per deployment tier; it is not enabled by default in every tier.

    Limitations
    • · “Device trust” here means registration, approval, revocation and cryptographic binding — configured per deployment, not a fixed always-on posture.

Isolation & authorization

  • Role-based access & tenant isolationImplemented

    Authorization is enforced server-side as permission checks on each route — not by hiding interface elements. A closed role-to-permission model applies least privilege, institution/tenant data is isolated at the data and permission layer, and access denials are logged. Cross-tenant access by platform or auditor roles is explicitly recorded in the audit trail.

    Scope: Roles include lab operator, authorized reviewer, institutional administrator and read-only auditor.

  • Environment separationImplemented

    Development, staging and production run in separate cloud accounts with independent data stores, storage and distributions. Code-level guards forbid development-only modes (demo seeds, mock runners, auto-trusted devices) from ever running in production and detect cross-environment resource mismatches.

    Scope: Applies across the platform.

Data protection

  • Encryption in transitImplemented

    Public traffic is served over HTTPS only (HTTP is redirected), using a modern TLS profile (TLS 1.2 or higher) with a managed certificate, and HTTP Strict Transport Security is enforced in production.

    Scope: Applies to the public site and the application API.

  • Encryption at restImplemented

    Application data stores and object storage are encrypted at rest with managed keys, the database is encrypted, and server-side key material is held in a managed key or secret store — never in plaintext on disk or in browser storage.

    Scope: Applies to managed application data and object storage.

  • Private storage & originImplemented

    Storage buckets block all public access and are reachable only through the content delivery layer with an origin access control; the database is not publicly accessible. Object storage is never browsable from the internet.

    Scope: Applies to web and application storage.

  • Secrets managementImplemented

    Workload secrets are supplied from a managed secret store at start-up, never written to logs, and the service fails fast if a secret is missing or a placeholder. The repository carries automated secret scanning and pre-commit checks, and published artifacts contain no secrets.

    Scope: Managed secret injection applies to staging and production workloads; rotation follows a documented procedure with preflight tooling.

    Limitations
    • · Scheduled automatic secret rotation is procedure-driven, not yet a proven automated mechanism.
  • The website is not a data channelImplemented

    The public website and ordinary email are not authorised channels for samples, sequences, genomic files, sample identifiers or patient data. Any evaluation involving data is conducted through a channel and conditions agreed with VBS.

    Scope: Applies to all public contact surfaces.

Execution integrity & provenance

  • Execution provenanceImplemented

    Each run preserves the technical context needed for review and reproducibility: a run identifier and timestamps, environment and pipeline/app version, container image digest, workflow hashes, result-contract and database/reference versions, input and output artifact hashes and status, the initiating actor and device, execution status and warnings, and an explicit register of any field that could not be determined — recorded as missing, never fabricated.

    Scope: Applies to every analysis run.

    Limitations
    • · Provenance supports reproducibility, review and auditability; it is not a clinical or regulatory sign-off.
  • Tamper-evident audit trailImplemented

    Security-relevant events pass through a single validated writer and are recorded as an append-only, per-tenant cryptographic hash-chain, so any modification, deletion or reordering is detectable by an independent verifier. Event types and metadata are restricted to a non-personal allow-list, and reads are limited by role.

    Scope: Applies to security-relevant events.

    Limitations
    • · The trail is tamper-evident: any modification is detectable by an independent verifier, rather than being prevented by the application alone.

Operation

  • Logging, health checks & alertsImplemented

    Health and liveness endpoints, structured request-correlated logging with a role-gated internal view that redacts personal data, and a catalogue of technical alerts mapped to on-call playbooks.

    Scope: Technical monitoring; alert-driven with on-call runbooks.

    Limitations
    • · There is no round-the-clock staffed security team; coverage is alert-driven with on-call procedures for a pre-clinical service.
  • Backups & recoveryImplemented

    The managed database has automated backups with a defined retention window and is not publicly accessible. Restore procedures are documented and periodically exercised with a non-destructive dry-run.

    Scope: The backup mechanism is implemented; continuous, signed-off restore verification is still being validated.

    Limitations
    • · Restores are drilled with a dry-run, not continuously verified end-to-end.
  • Incident-response proceduresBeing validated

    A maintained set of severity-tagged incident-response runbooks (pipeline, API, database, storage, secrets rotation, threat findings, deploy rollback), validated for structure in continuous integration, together with automated remediation for certain cloud threat findings.

    Scope: Runbooks are documented and CI-validated; end-to-end incident exercises are pending.

    Limitations
    • · Not yet exercise/tabletop-proven end to end.
  • Web security headersImplemented

    The public site sends hardening response headers — HTTP Strict Transport Security, anti-framing, no MIME sniffing and a prudent referrer policy — and the staging environment is marked non-indexable.

    Scope: Applies to the public website.

  • Content Security PolicyBeing validated

    A Content Security Policy that restricts where scripts, styles, fonts, images and connections may load from is being rolled out — applied and verified on staging first, then production.

    Scope: Applies to the public website; production rollout after staging verification.

    Limitations
    • · On the static host the framework's inline bootstrap script currently requires an inline-script allowance; a hash-based tightening is planned.

Deployment

  • VBS-operated managed cloudImplemented

    NextGenAMR currently runs as a VBS-operated managed cloud service on AWS, with the operational responsibility, updates, secret handling, logging and backups managed by VBS. Other deployment models are not offered as generic options; the deployment design, responsibilities and data handling are reviewed during an institutional evaluation.

    Scope: The only current deployment modality.

    Limitations
    • · Dedicated / single-tenant / private-cloud / on-premise / hybrid models are not currently available and are assessed only during an institutional evaluation.

What data the website handles

The public website and ordinary email are not authorised channels for samples, sequences, genomic files, sample identifiers or patient data. Any evaluation involving data uses a channel and conditions agreed with VBS. Product data (processed within the platform) is distinct from the limited contact information you may send through this site.

Each run preserves provenance and execution metadata and produces structured, auditable outputs for professional review (JSON and PDF).

How data flows (simplified)

Illustrative

A functional, illustrative view of the zones an authorised analysis passes through — not a network topology. No addresses, resource names or internal relationships are shown.

  1. 01Authorised input

    A supported input received through a controlled channel agreed with VBS — never the public website.

  2. 02Processing layer

    The VBS-operated cloud runs the versioned bioinformatics workflow within scope.

  3. 03Evidence & provenance

    Structured evidence and per-run provenance are generated for review and reproducibility.

  4. 04Controlled persistence

    Results and artifacts are stored encrypted, with role-based access and an audit trail.

  5. 05Authorised review

    Authorised professionals review the outputs together with other applicable tests.

  6. 06Reviewable outputs

    Structured outputs and provenance for professional review — never an autonomous clinical decision.

Documentation available for institutional review

Each item carries a status and how it can be shared. Where a document is not yet published, requesting a security brief opens the institutional conversation.

  • Security brief (summary)Available

    A concise public summary of the security and deployment approach.

    Public

  • Simplified data-flow diagramAvailable

    A functional, illustrative view of authorised input, processing, provenance, review and outputs.

    Public

  • Architecture overviewBeing validated

    A high-level architecture overview for institutional review.

    On institutional request

  • Access-control summaryBeing validated

    A summary of identity, roles and access controls, without internal matrices.

    On institutional request

  • Backup & recovery summaryBeing validated

    A description of backup and recovery procedures at a reviewable level.

    On institutional request

  • Subprocessors overviewBeing validated

    The current list of subprocessors relevant to the service.

    On institutional request

  • Incident-response procedureBeing validated

    An overview of the incident-response procedure, shared under a confidentiality agreement.

    Under NDA

  • Due-diligence questionnaire responseAvailable

    VBS can respond to institutional security and due-diligence questionnaires.

    On institutional request

Request a security brief

Reporting a security issue

Security reports are currently received through the general Vanguard Biotech Systems contact address. This is the company's current contact channel, not a dedicated security operations team. To help us route it, please include “Security report” in the subject.

vanguardbiotechsystems@gmail.com

  • Describe the issue, the observed impact and reproducible steps; contact details are optional.
  • Please do not access, modify, delete or exfiltrate data, and do not run denial-of-service tests or social engineering.
  • Please allow a reasonable review period before publishing any details.

This is not a paid bug-bounty programme; we do not currently offer rewards or a fixed response time.

See /.well-known/security.txt

Questions institutions ask

Where is NextGenAMR deployed today?

NextGenAMR currently runs as a VBS-operated managed cloud service on AWS, with the operational responsibility, updates, secret handling, logging and backups managed by VBS. Other deployment models are not offered as generic options; the deployment design, responsibilities and data handling are reviewed during an institutional evaluation.

Can we send data through the website?

The public website and ordinary email are not authorised channels for samples, sequences, genomic files, sample identifiers or patient data. Any evaluation involving data uses a channel and conditions agreed with VBS. Product data (processed within the platform) is distinct from the limited contact information you may send through this site.

Does security mean the product is clinically validated?

Security controls (access, encryption, logging, provenance) are not clinical validation, a medical-device conformity mark, or authorisation for diagnostic use.

What does a run record keep?

Each run preserves provenance and execution metadata and produces structured, auditable outputs for professional review (JSON and PDF).

How do we report a vulnerability?

Security reports are currently received through the general Vanguard Biotech Systems contact address. This is the company's current contact channel, not a dedicated security operations team. To help us route it, please include “Security report” in the subject.

Institutional security review

Vanguard Biotech Systems does not currently make regulatory-authorisation or autonomous-clinical-use claims for NextGenAMR. Intended use and regulatory strategy will be assessed as the product evolves.