Medical AI Guides

In-depth Medical AI guides for research, medtech and medical-AI teams. Methodical, sourced and defensible practice.

11 articles

Medical AI

How to Write the Statistics Section of a Clinical Evaluation Report (CER)

How to write the CER statistical core: MEDDEV 2.7/1 Rev 4, EU-MDR Annex XIV, and the biostatistician's role in state-of-the-art appraisal and data synthesis. It is not a QMS document, it is a clinical evidence analysis.

June 30, 2026 · 8 min read

Medical AI

EU AI Act Article 10: Data Governance and Representativeness in Medical AI

EU AI Act Article 10 defines data governance, dataset representativeness, and bias obligations for high-risk medical AI. A subgroup analysis and data quality plan ahead of the August 2026 entry into force.

June 30, 2026 · 8 min read

Medical AI

RWE (Real-World Evidence) Medical-AI Protocol

Real-world evidence (RWE) for medical AI: study design, protocol, SAP, confounding and bias control, the FDA RWE guidance (December 2025), and how to set up post-market surveillance.

June 30, 2026 · 9 min read

Medical AI

MRMC Multi-Reader Study: Proving AI Diagnostic Performance

Proving AI diagnostic performance with an MRMC multi-reader multi-case study: ROC/AUC, reader variability, AI-vs-reader, AI-assisted reading, and FDA expectations.

June 30, 2026 · 9 min read

Medical AI

Is Our SaMD Class IIa or IIb? Software Medical Device Classification

Medical-AI / SaMD Class IIa or IIb? How MDR Annex VIII Rule 11, MDCG 2019-11, and the intended use statement determine the class: the surgical intervention and serious deterioration fault lines, a classification decision guide.

June 30, 2026 · 6 min read

Medical AI

How to Design a SaMD Clinical Validation Study

How to design a SaMD clinical validation study: standalone performance, study design, sample size, AUC-ROC, reference standard, internal vs external validation guide.

June 30, 2026 · 9 min read

Medical AI

How to Prepare a TITCK CDSS Clinical Validation Report

What a TITCK CDSS/SaMD clinical validation report must contain and who prepares it: EU-MDR transposition, standalone performance, training-data representativeness, risk management, and the statistical core.

June 30, 2026 · 8 min read

Medical AI

TRIPOD-AI and PROBAST-AI: Reporting AI Diagnostic Models

How to report an AI diagnostic/prognostic model with TRIPOD-AI, how to assess risk-of-bias with PROBAST-AI, and the common gaps that reviewers and regulators look for.

June 30, 2026 · 8 min read

Medical AI

Data Leakage and Train-Test-Split Errors in AI Validation

How data leakage and train-test-split errors sink a medical AI validation file: patient-level split, external validation, and the FDA independent-test expectation.

June 30, 2026 · 8 min read

Medical AI

Clinical AI Validation Pipeline: TRIPOD/DECIDE-AI

The path of medical AI models to clinical use: internal validation, external validation, prospective deployment study, TRIPOD-AI 2024 + DECIDE-AI reporting, ROC vs decision curve analysis, calibration, ethics committee requirements, and the TITCK SaMD process.

May 26, 2026 · 9 min read

Medical AI

Artificial Intelligence in Radiology: nnU-Net Segmentation

For teams that want to develop an AI model in medical imaging: the nnU-Net architecture, the 2D vs 3D decision matrix, transfer learning, validation strategy, TRIPOD-AI compliant reporting, and a clinical validation roadmap. A practical technical guide for medical artificial intelligence.

May 23, 2026 · 10 min read