Software as a Medical Device: How to Pass the Intended Use Test

What defines software as a medical device?

Whether your software qualifies as a medical device depends entirely on its intended use, according to guidance from October 8, 2026 [1, 2]. If your tool makes diagnostic or treatment claims, regulators treat it differently than a general wellness app. This distinction shifts your legal obligations overnight. It turns a standard software project into a regulated medical product. You must audit your marketing materials and product functions against these definitions. For developers, ignoring this categorization invites significant regulatory risk. The key is analyzing your specific intent, not just your underlying technology. If your software claims to diagnose, treat, or prevent disease, the regulatory burden applies regardless of how you built the tool [1, 2].

How does the FDA intended use guidance impact your product?

On October 7, 2026, Ōura earned ISO 13485 certification for its medical device software [3]. This move signals that the company has implemented a formal quality management system to meet international standards for medical products. According to reports from October 8, 2026, this certification is a benchmark for brands seeking to bridge the gap between consumer tech and clinical-grade tools [2]. It is not just about having functional code; it is about proving the processes behind that code ensure user safety. This industry shift reinforces that regulators are watching how apps characterize their own data.

How to Manage Digital Health Regulatory Risk as a Developer

The core issue is the specific claim your software makes to the user [1]. If your AI tool assists in clinical decision-making or monitors a chronic condition, it likely falls under medical device regulation. However, if it only tracks general fitness metrics, it often remains outside that scope [2]. Regulators look at what you tell the user, not just what the software does under the hood. If you promise a specific health outcome, you are likely selling a medical device. This makes your marketing copy a primary document for regulatory scrutiny.

Steps for medical device software compliance

This affects anyone building software that touches health outcomes [2]. It applies to AI startups aiming for clinical diagnostic tools, as well as established apps pivoting into health monitoring [1]. If you are a developer, your marketing copy now carries significant legal weight. Even a minor claim about detecting a condition can trigger a mandatory classification change. You cannot treat your app store description as a casual sales pitch.

The Hidden Costs of Compliance

Compliance is expensive and time-consuming. Obtaining certifications like ISO 13485 requires a large financial investment and rigorous internal oversight [3]. You must maintain detailed documentation for every update, track every code change, and undergo regular third-party audits. These costs can easily exceed your initial development budget. A major downside exists: the faster you want to ship new features, the harder it is to stay compliant. Your development speed will likely slow down to accommodate these necessary safety checks.

What Should You Do Next

Verify your current claims against your regulatory roadmap. Check whether your software’s technical documentation matches your public-facing marketing [1]. If you are uncertain, consult a specialist who understands medical device classification. Waiting until a regulator flags your app is a strategy that rarely succeeds. You should also audit your user interface to ensure no unintended diagnostic suggestions appear in your output.

What Is Still Unknown

It remains unclear how regulators will handle generative AI features in the future [1]. While static algorithms have clear rules, AI that adapts over time creates an evolving target for auditors. We do not yet know how individual countries will harmonize these rules for global software products. Until clear standards emerge for self-learning systems, you should assume the strictest interpretation applies.