UK CERSI 2026 · THEME 1: DATA-DRIVEN APPROACHES AND AI
Tanisha TejadaThoreau
Regulatory science for auditable clinical AI
I am seeking to contribute to a UK consortium bidding for the MHRA, MRC and Office for Life Sciences Centres of Excellence in Regulatory Science and Innovation, as an international co-lead, work-package contributor or advisory board member.
RESEARCH FOCUS
Making continuously learning AI auditable
I am a regulatory scientist, quality systems auditor and JD candidate with more than 15 years across analytical chemistry, biopharmaceutical operations, clinical research quality and health law. My work asks a practical question: how can AI systems that keep learning be audited to the standard regulators already expect of laboratory and clinical data?
To answer it I developed the Algorithmic Audit Doctrine (AAD), a four-pillar framework for AI in regulated settings. In September 2026 I submitted a public comment to the US FDA on the regulation of generative AI-enabled medical devices (Docket FDA-2026-N-7874).
The four pillars
- 1. The Algorithmic Audit: continuous, runtime governance of AI reasoning across its operational lifecycle
- 2. Algorithmic Competency Assurance: ISO/IEC 17025 § 6.2 proficiency testing, adapted to AI through blind control re-assessments
- 3. Algorithmic Materiality: a three-tier risk matrix with quantitative thresholds for drift and degradation
- 4. Algorithmic Chain of Custody (A-CoC): a tamper-evident audit trail for every AI-assisted result
FOR A THEME 1 CENTRE
Open research questions I would bring
01
What must continuous, runtime governance of AI reasoning include to replace one-off static validation in GxP settings?
02
How often, and against which blind control sets, should adaptive AI be re-assessed before it may keep processing live clinical trial data?
03
Which quantitative drift and performance thresholds should separate low, moderate and high materiality, and when should a system lock automatically?
04
Can a six-component Algorithmic Chain of Custody record meet both FDA and MHRA data-integrity expectations for AI-assisted results?
CONSORTIUM FIT
What I would contribute
Contribution
- Audit methods ready to test in pilot studies
- Hands-on inspection experience across GCP, CLIA/CAP and ISO 15189 / 17025
- A US–UK regulatory comparison for AI-enabled products
- Industry and laboratory perspective on what auditors actually need to see
Roles I am open to
- International co-lead
- Work-package contributor
- Advisory board member
Sample audit reports
IMPLEMENTATION TOOLKIT
Ready to implement, not just a framework
The AAD comes with the working documents needed to put it into practice inside a regulated organisation.
PROT-AAD-001 · PILLAR 1
The Algorithmic Audit
Moves quality assurance from one-off code verification to continuous, runtime governance of AI reasoning across its entire operational lifecycle.
PROT-AAD-002 · PILLAR 2
Algorithmic Competency Assurance
Adapts analyst proficiency testing (ISO/IEC 17025 § 6.2) to AI: periodic blind control re-assessments before a tool may keep processing live clinical trial data.
PROT-AAD-003 · PILLAR 3
Algorithmic Materiality
A three-tier risk matrix with quantitative thresholds for drift and degradation. Tier 3 events trigger immediate system lock and formal re-validation.
PROT-AAD-004 · PILLAR 4
Algorithmic Chain of Custody (A-CoC)
A tamper-evident audit trail for every AI-assisted result, with six mandatory components, from the model’s weight hash to the reviewing scientist’s e-signature.
DOS-AAD-001 · INSPECTION
Master Inspection Dossier
The consolidated evidence package an organisation presents to an inspector.
CHK-AAD-001 · FIELD USE
Field Audit Checklist
The on-site checklist an auditor works from during an AAD assessment.
The full toolkit is available to consortium partners on request. All six documents are drafted and ready to be tested in a pilot setting.
TRACK RECORD
Selected record and public comments
Selected record
- Led the greenfield launch and licensing of six analytical testing laboratories
- Authored ISO/IEC 17025 quality management systems used across dozens of facilities
- Global Clinical QA Lead Auditor, ACM Global Laboratories
- Served on the clinical team supporting COVID-19 remdesivir trials during Operation Warp Speed
- Contributor to AOAC CASP consensus standards development
Public comments
- US FDA, Considerations for the Regulation of Generative AI-Enabled Medical Devices. Docket FDA-2026-N-7874, Document ID FDA-2026-N-7874-0102, posted 21 September 2026.
- NIST, IR 8623 (initial public draft), CSF 2.0 Community Profile for Federal Agency O-RAN Deployment. Comment accepted.
EDUCATION AND CREDENTIALS
Education and credentials
Education
- JD Candidate, Taft Law School. Healthcare and clinical law, administrative procedure, data privacy and information governance
- MS Analytical Chemistry, Illinois Institute of Technology
- MS Biotechnology, University of Maryland Global Campus
- BS Biochemistry and Molecular Biology, University of Maryland, Baltimore County
- Practical Nursing Diploma (LPN), Platt Medical College
Credentials
- AI and information security: ISO/IEC 42001 Lead Auditor; ISO/IEC 27001
- Laboratory and device: Certified Lead ISO 17025 Auditor; ISO 15189, ISO 13485, ISO 14971, ISO 9001
- Clinical: Advanced ICH GCP (AGCPC); Clinical Research Associate (CRA); 21 CFR Parts 11, 50, 54, 56, 312; Joint Commission / CMS CoP
- Rulemaking and operations: Rulemaking and Regulatory Monitoring training; graduate coursework in advanced project management and Lean Six Sigma
CONTACT
Discuss a consortium bid
If you are leading or joining a Theme 1 bid, I would welcome a conversation about where the AAD and my audit experience could strengthen it. Applications close 12 January 2027.