In support of digital health product innovation, the Health Sciences Authority (HSA), Singapore, collaborated with the Ministry of Food and Drug Safety (MFDS), Korea, to publish guiding principles for conducting clinical trials of machine learning-enabled medical devices (MLMD). This guideline version is effective from 11 December 2024.
Core Guiding Principles for MLMD Clinical Trials
- Clinical Trial Design
Alignment to research objectives and the product’s intended use is required to generate robust scientific clinical evidence. Supplementary studies may be included where applicable. - Patient and Test Dataset Selection
Must be independent from the training dataset. Establish clear inclusion and exclusion criteria for study participants to ensure valid trial results. - Selection of Clinical Reference Standard and Data Interpretation
Use current recognized clinical guidelines where available. For novel use cases, clinical and domain experts should establish a reference guideline, with rationale for consensus documented. - Primary Endpoint and Results Analysis
May be based on proposed performance indicators (non-exclusive). Statistical methods should be used for analysis, measured against acceptance criteria.
Additionally, clinical trials should adhere to all applicable local laws and regulations.
View full guideline here:
https://www.hsa.gov.sg/announcements/nex2us-newsletter/nex2us-newsletter-issue-11—may-2025
Implications to Clients
Clients developing machine learning-enabled medical devices (MLMD) should align trial design, data selection, and analysis methods with these principles to ensure regulatory compliance and the generation of robust clinical evidence.
For more information or assistance, contact us at sales@andamanmed.com or click the button below.
