What is FDA Software as a Medical Device (SaMD)? Definition, Formula, and Benchmark

FDA Software as a Medical Device (SaMD) is the regulatory category defining standalone software that performs one or more medical functions — diagnostic, therapeutic, monitoring, or clinical decision support — without being part of a hardware medical device. SaMD encompasses a broad range of health software including AI diagnostic tools, clinical decision support systems, remote patient-monitoring software, and certain digital therapeutics.

FDA classification determines regulatory rigor. Class I (low risk) SaMD generally requires registration and listing but no premarket authorization. Class II (moderate risk) typically requires 510(k) premarket notification. Class III (high risk) requires Premarket Approval (PMA) with extensive clinical evidence. Healthcare AI systems span all three classes depending on intended use and risk.

The 21st Century Cures Act clarified that certain clinical-decision-support software is outside FDA SaMD scope when it meets specific criteria: it informs clinicians rather than replacing their judgment, displays sources such that clinicians can independently evaluate recommendations, and is not used to analyze medical imaging or signals. This "CDS carve-out" enables many healthcare AI systems to operate outside formal FDA SaMD pathway while still being regulated by other health-oversight mechanisms.

Predetermined Change Control Plan (PCCP) is a newer FDA pathway specifically for AI/ML-based SaMD. It addresses the traditional pathway's difficulty with AI/ML systems that evolve post-approval. Under PCCP, sponsors can specify pre-approved modification parameters; subsequent changes within those parameters do not require new premarket authorization. This framework enables AI systems to learn and update without repeatedly pausing for FDA review.

For healthcare AI development and deployment, SaMD classification is a first-order consideration. Products intended for diagnostic or therapeutic use must navigate FDA pathway; products intended for administrative, revenue-cycle, or operational use typically do not. The boundary between SaMD and non-SaMD is sometimes ambiguous and benefits from regulatory-counsel engagement for new product categories.

International regulatory frameworks (EU MDR, UK MHRA, Canada Medical Devices Regulations) have similar concepts with varying specifics. Global deployments of healthcare AI must consider each jurisdiction's requirements, which can significantly affect product strategy.

For RCM-focused AI (denial management, eligibility verification, coding, documentation), most applications are non-SaMD because they do not make diagnostic or therapeutic decisions. Use cases that cross into clinical decision support (risk stratification driving care decisions, predictive algorithms affecting treatment selection) enter SaMD territory and require regulatory analysis.

From a board-reporting standpoint, FDA Software as a Medical Device (SaMD) belongs in the compliance committee's quarterly dashboard. The reporting line should include volume, exception rate, and any open remediation action; reviewers tie FDA Software as a Medical Device (SaMD) metrics to the broader compliance program KPIs so an emerging FDA Software as a Medical Device (SaMD) risk surfaces before it becomes a formal finding. Pairing the FDA Software as a Medical Device (SaMD) trend with model validation clinical gives the committee a single view of whether the control environment is strengthening or drifting.