Overview - ISO/TR 24291:2021 (Health informatics | Machine learning in medicine)
ISO/TR 24291:2021 is an ISO Technical Report that catalogs and defines categories of clinical use cases for machine learning (ML) technologies in medical imaging and other medical applications. The report focuses on the clinical side - the ways ML is applied in patient care - and defines clinical usages and necessities for ML-based solutions. It explicitly excludes upstream activities (data collection/curation, pre‑processing, model training and validation), non‑ML AI methods (e.g., symbolic AI, expert systems), basic research, and non‑human/veterinary use.
Key topics and requirements
- Use‑case categorization: organizes ML applications by
- Technology (e.g., deep learning, image processing, natural language processing, audio recognition, robotics, continuous monitoring, big‑data analysis, prediction modeling)
- Medical specialty (e.g., radiology, pathology, dermatology, ophthalmology, internal medicine, cardiology, neurology, surgery, anaesthesia/ICU, emergency)
- Clinical usage (e.g., clinical trials, clinical assistance, precision medicine, imaging/diagnostics, hospital management, robotic surgery, drug development)
- Definitions and terminology: standardized terms such as AI, ML, deep learning, CADe/CADx, CDSS, EMR, etc., to support consistent communication across stakeholders.
- Scope constraints: clarifies what is and is not covered so implementers know the document is for clinical application categories and clinical requirements - not for data engineering or ML model training processes.
- Practical orientation: emphasizes clinical scenarios requiring repeated detection/diagnosis, real‑time monitoring, and treatment prediction with images and continuous signals.
Applications - who uses ISO/TR 24291:2021 and why
- Health IT and medical device companies: to map product functionality to recognized clinical use‑case categories and identify gaps for new ML applications.
- Clinicians and hospital IT: to understand where ML can support real‑time monitoring, diagnostic assistance, treatment prediction, and workflow automation.
- Regulators and standards developers: to reference standardized clinical categories when assessing conformity, safety, and interoperability.
- Researchers and product managers: to prioritize clinical deployment scenarios that align with documented clinical needs.
Practical application examples include ML for lung cancer screening/reporting, chest CT analysis, clinical decision support systems, personalized drug therapy optimization, and assistive speech technologies.
Related standards
ISO/TR 24291 references and aligns with existing ISO/IEC terminology and health informatics guidance (for example, ISO/IEC TR 29119‑11, ISO/IEC 20546, ISO/TS 22756). Users should consult these and other relevant standards for complementary requirements (terminology, big data handling, clinical decision support).
By defining clear clinical categories and terminology, ISO/TR 24291:2021 helps accelerate safe, interoperable adoption of machine learning in medical imaging and broader clinical practice.