In comments on a federal RFI, the association says laboratories should retain responsibility for validating, implementing, and monitoring AI tools used in patient testing.
The Association for Diagnostics & Laboratory Medicine (ADLM) has submitted comments in response to a federal request for information regarding updates to the Clinical Laboratory Improvement Amendments (CLIA), recommending that artificial intelligence (AI) tools remain subject to standard clinical laboratory oversight.
CLIA regulations govern all clinical laboratory testing in the US, but the framework has not been modernized to address AI-driven testing tools since its implementation in 1992, the association noted. In its comment letter, ADLM recommended that emerging AI and machine learning systems be held to the same professional expertise, quality systems, validation requirements, and monitoring standards that apply to conventional laboratory tests.
“AI has the potential to support tremendous advances in laboratory medicine, but innovation in this area must be balanced with the need to ensure test quality and patient safety,” says Dr Stanley F. Lo, ADLM president, in a release. “Laboratories have the expertise and quality systems needed to evaluate, implement, and continuously monitor AI tools, making continued laboratory oversight essential to the responsible use of AI in clinical laboratory testing.”
Addressing Unique Failure Modes in Laboratory AI
While traditional software has supported functions such as verifying, interpreting, and reporting results, emerging AI-based applications introduce distinct safety challenges not covered by current CLIA standards, according to the release.
ADLM pointed out that an error in conventional software typically affects every patient case matching the same programmed conditions, making it straightforward to assess using cases with known expected outputs. In contrast, AI models can produce case-specific errors that are more difficult to isolate and troubleshoot. Furthermore, generative AI tools can yield inaccurate or unsupported information, omit clinically important details, or change behavior following updates to the model, prompt, or knowledge base, the association stated.
Focusing on the Total Testing Process
Rather than creating a separate regulatory structure for AI as standalone software, ADLM emphasized that these technologies should be regulated within CLIA’s existing total testing process framework, according to the release.
The association stated that targeted updates to CLIA should reflect the unique challenges of machine learning while ensuring that clinical laboratorians maintain responsibility for the validation, implementation, and routine monitoring of AI tools in patient care.
ID 465664072 © Gaewchiangmai | Dreamstime.com
