Podrobnosti návrhu
| Název | Testing of Artificial Intelligence / Machine Learning-enabled Medical Devices |
|---|---|
| Číslo | 62/582/CDV - IEC 63450 ED1 |
| Zdroj | ISO\IEC\CEN\CENELEC |
| Komise | IEC/TC 62 (Medical equipment, software, and systems) |
| Zkontrolujte datum zahájení | |
| K připomínkám do | |
| Oblast zaměření |
Akce
Anotace
Software testing is well treated in standardization (see ISO/IEC/IEEE 29119-1:2022 [1]). For medical devices, requirements on software testing are well established internationally by IEC 62304:2006 [2] and IEC 62304:2006/Amd 1:2015 [3]. This document establishes additional aspects to specifically test AI components (3.2) in AI/ML-enabled medical devices (3.4).
Al components (3.2) considered in this document include AI models (3.3) without machine learning, locked (3.17) ML models (3.20) and ML models (3.20) that learn under the control of the manufacturer.
Additional requirements for the case where an ML model (3.20) is continually learning in the field under indirect control (3.34.2) by the manufacturer are in Annex A.
Tests can be conducted at various lifecycle stages of the AI/ML-enabled medical device (3.4), depending on the nature of the AI model (3.3) .
Instrukce
Tento dokument můžete připomínkovat po jednotlivých částech - stačí u příslušné části dokumentu vyplnit formulář v souladu s pokyny a kliknout na ‚Odeslat připomínku‘.
Prosíme, nepoužívejte vulgární výrazy a nevyužívejte tento prostor pro umisťování reklamy.