Skip to main content
Ethics

Ethics Committee Response Kit

Template responses for the most common questions raised by Human Research Ethics Committees (HRECs), institutional review boards, and university ethics offices when a researcher proposes to use a commercial clinical decision support tool. Copy and customise these responses for your ethics application.

Copy responses into your ethics application
Print (Ctrl+P / Cmd+P) to save as PDF

Is this tool validated for clinical use? What is its evidence base?

Typical ethics committee question

Template response — copy and customise for your ethics application

OpenBook Clinical is a clinical decision support (CDS) tool, not a diagnostic or therapeutic medical device. It retrieves and synthesises peer-reviewed, published literature to support evidence-based practice. It does not replace clinical judgment, diagnosis, or treatment decisions.

The platform's evidence retrieval function operates over indexed published literature and surfaces citations with NHMRC evidence grades (Level I–IV) where applicable. All AI-generated summaries include the source citations that generated them, allowing the clinician or researcher to verify the original evidence directly.

Consistent with NHMRC guidance on evidence hierarchies, researchers using the platform are expected to apply critical appraisal to AI-generated summaries. The platform is not intended to serve as the sole basis for clinical or research conclusions. Its role is to accelerate literature discovery and reduce the burden of initial evidence synthesis, not to replace systematic review methodology.

A detailed description of the platform's limitations, model classification, and known performance characteristics is published in the AI Governance Framework at openbookclinical.com.au/governance/ai-framework.

Note for applicant: You may wish to add a sentence describing how you will use the tool in your specific study (e.g., 'OpenBook Clinical will be used to support rapid literature review during clinical rotations as part of [Study X]') and how findings will be cross-validated.

How do you ensure no patient data enters the system? What technical controls prevent this?

Typical ethics committee question

Template response — copy and customise for your ethics application

OpenBook Clinical is designed by architecture to prevent patient data from being entered. Specifically:

  • There are no patient-record fields in the user interface. The platform does not have fields for patient name, date of birth, Medicare number, address, or clinical notes. Users cannot create a patient record.
  • The platform implements real-time PII detection that identifies Australian personally identifiable information patterns (including Medicare number format, name + DOB combinations, and other identifiers) in search queries and displays an immediate warning to the user.
  • The platform's Terms of Service explicitly prohibit the entry of patient-identifiable information, constituting a contractual obligation on all users.

In our research, we will additionally provide all participating clinicians / researchers with explicit written instructions not to enter patient-identifiable information into the platform, and will include this obligation in our participant information sheet and consent form.

Note for applicant: Add a sentence about any additional procedural controls your team will use (e.g., training, monitoring, research protocols).

What happens if the AI gives incorrect or misleading information? How is this managed?

Typical ethics committee question

Template response — copy and customise for your ethics application

OpenBook Clinical is a decision-support tool, and all outputs require verification by the clinician or researcher before acting upon them. The platform implements several controls to support accuracy and appropriate use:

  • Citation integrity: Every AI-generated response includes the source citations that informed it. Researchers are expected to — and are instructed to — verify claims against the primary literature.
  • Limitations disclosure: The platform displays a prominent disclaimer with every response noting that AI-generated summaries may contain errors, may not reflect the most current evidence, and must be reviewed by a qualified clinician before any clinical application.
  • AHPRA professional obligations: All registered users of the platform are AHPRA-registered professionals (or researchers supervised by them) who retain their professional obligations to exercise independent clinical judgment. The platform does not and cannot override professional responsibility.
  • No autonomous action: The platform provides information only. It does not order tests, prescribe medications, or initiate care. Every output is reviewed by a human before any action is taken.

In our study, all AI-generated outputs will be reviewed by [insert: researcher / supervising clinician] before any research conclusions are drawn. We will record instances where AI output was found to be inaccurate or incomplete as part of our study methodology.

Note for applicant: Customise the last paragraph to describe your specific research protocol for handling AI outputs, including how errors or discrepancies will be documented.

Can you provide a security certificate or evidence of security assessment for this vendor?

Typical ethics committee question

Template response — copy and customise for your ethics application

OpenBook Clinical publishes the following security documentation, all of which can be referenced in your ethics application or provided to your IT department:

  • Trust & Security Centre — published at openbookclinical.com.au/trust. Covers infrastructure, encryption, access controls, and operational security.
  • Essential Eight Maturity Self-Assessment — published at openbookclinical.com.au/security/essential-eight. Aligned with the Australian Signals Directorate (ASD) Essential Eight framework.
  • Pre-filled Vendor Security Questionnaire — published at openbookclinical.com.au/governance/research/vendor-questionnaire. Suitable for submission to your institutional IT department.
  • Sub-processor certifications: Primary sub-processors (Supabase, Anthropic, Vercel, Postmark) hold SOC 2 Type II certification. Payment processor (Stripe) holds PCI DSS Level 1 certification. Details at openbookclinical.com.au/governance/sub-processors.

An independent penetration test is scheduled for 2025. Results will be made available to institutional assessors under NDA upon request at openbookclinical@gmail.com.

Note for applicant: If your institution requires additional security documentation (e.g., signed security questionnaire, architecture diagram), contact openbookclinical@gmail.com. We typically respond within 3 business days.

Who is legally responsible if the tool provides harmful advice that influences a clinical or research decision?

Typical ethics committee question

Template response — copy and customise for your ethics application

OpenBook Clinical's legal position — which is consistent with standard clinical decision support software — is that the platform is a tool to assist professional judgment, not to replace it. Accordingly:

  • Professional responsibility: The clinician or researcher using the platform retains all professional obligations under their AHPRA registration. The duty of care to patients (in clinical contexts) and the responsibility for research integrity (in research contexts) remains with the practitioner, not the software vendor.
  • Platform limitations notice: OpenBook Clinical's Terms of Service and in-platform disclosures make clear that outputs are informational only, may contain errors, and must be verified before clinical or research application. Users agree to these terms upon registration.
  • No clinical decision-making: The platform does not diagnose, prescribe, or direct care. It surfaces published evidence. The researcher or clinician makes all decisions based on their professional judgment, institutional protocols, and the totality of available evidence — not solely on platform output.
  • Institutional indemnity: Researchers using the platform within the scope of their institution's approved research protocol are typically covered by their institution's professional indemnity and research insurance arrangements. We recommend confirming this with your institution's risk management team.

This position is consistent with the use of other reference tools (UpToDate, clinical guidelines, textbooks) in research and clinical practice, none of which transfer professional liability from the practitioner to the publisher.

Note for applicant: You may wish to confirm the indemnity position with your institution's legal or risk team, particularly if your research involves clinical interventions.

Is this a registered medical device (Software as a Medical Device / SaMD)? Does it require TGA approval?

Typical ethics committee question

Template response — copy and customise for your ethics application

OpenBook Clinical is not a registered medical device and does not require TGA approval, based on the following analysis:

  • The platform is a clinical decision support (CDS) tool in the category of “general purpose software used in healthcare settings.” It retrieves and synthesises published literature to support professional decision-making.
  • Consistent with TGA guidance on Software as a Medical Device (SaMD), CDS tools that do not perform automated diagnosis, treatment selection, or autonomous clinical action are generally excluded from the definition of a medical device under the Therapeutic Goods Act 1989 (Cth) where a clinician reviews and independently acts on the information provided.
  • OpenBook Clinical does not diagnose conditions, recommend specific treatments, calculate drug doses, or make autonomous clinical decisions. All outputs are reviewed and acted upon by a qualified clinician.

OpenBook Clinical's detailed TGA classification rationale is available on request at openbookclinical@gmail.com. Researchers are encouraged to confirm this classification position with their institution's research ethics office if the platform is to be used in a clinical research context.

Note for applicant: If your ethics committee raises specific concerns about TGA classification, contact openbookclinical@gmail.com for a detailed written classification rationale you can submit alongside your application.

Will data from the tool be published? Can we include usage data or AI outputs in research publications?

Typical ethics committee question

Template response — copy and customise for your ethics application

Researchers may reference, quote, and publish AI-generated outputs from OpenBook Clinical in research publications, subject to the following:

  • Attribution: Publications should acknowledge the use of OpenBook Clinical as a tool in the evidence retrieval process, consistent with standard practice for reporting research methods (PRISMA, CONSORT, or equivalent).
  • Citation integrity: AI-generated summaries must be verified against the cited primary sources before citation in a research publication. The AI output itself should not be cited as a primary source.
  • Account data: Usage metadata (search logs, session data) is not intended for publication. If you wish to include platform usage data in your research (e.g., query logs as a study outcome), please contact us at openbookclinical@gmail.com to discuss a data sharing arrangement.
  • Screenshots: Screenshots of the platform interface may be included in publications with attribution to OpenBook Clinical.

Note for applicant: If your research design specifically involves analysing OpenBook Clinical usage data as a study outcome, a data sharing agreement may be required. Contact openbookclinical@gmail.com early in your project planning.

Question not covered here?

If your ethics committee raises a question not covered above, contact our research team at openbookclinical@gmail.com. We can provide written responses suitable for inclusion in your ethics application within 3 business days.

OpenBook Clinical Pty Ltd · ABN 38 698 494 656 · Sydney, NSW, Australia