iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Artificial intelligence is already used in healthcare to interpret medical images and signals, support diagnosis and monitoring, guide some treatment functions, and help with drug development, disease surveillance, outbreak response, and health-system management. But an AI-enabled device’s FDA authorization is tied to that device’s intended use; it is not proof that AI improves every patient outcome or is suitable for every clinical setting.
What is AI doing in healthcare now?
“AI in healthcare” covers different kinds of work, not one technology or a single level of clinical responsibility. The World Health Organization (WHO) describes applications across diagnosis and clinical care, drug development, disease surveillance, outbreak response, and health-systems management. Some tools support a clinician’s assessment; others may be designed to carry out a defined function within a treatment process. The task, user, setting, and evidence matter as much as the label “AI.”
Diagnosis and clinical assessment
Medical-device examples described by the U.S. Food and Drug Administration (FDA) include software that provides diagnostic information about skin cancer and tools that detect diabetic retinopathy from retinal images. These are examples of intended functions—not, by themselves, evidence that a tool improves outcomes across all patients, clinics, or uses.
Imaging and signal processing
Some AI-enabled devices process images or physiological signals. An algorithm may sharpen an image to support interpretation, for example, or analyze a signal to estimate a health risk. The FDA also describes a sensor device that estimates the probability of a heart attack. A device’s output should be understood in the context of its specific intended use and the role of the clinician or other user.
#1 Best Overall
Monitoring and treatment functions
Some devices are designed to use continuous glucose monitor readings to automate insulin dosing. This is a treatment-related function, rather than simply an image-classification task. Its presence on an FDA device list does not establish that every automated dosing system, configuration, or care setting has the same performance or benefit.
Work beyond an individual clinical encounter
WHO’s broader map includes drug development, disease surveillance, outbreak response, and health-system management. These applications can involve different data, users, and decisions from those of a medical device used to assess one patient. They should not be assumed to fall under the same regulatory pathway simply because they use AI.
What does the FDA authorize, and what does that mean?
The FDA’s key distinction is: “The FDA does not regulate AI as such; it regulates medical devices, including AI-enabled medical devices.” In the United States, FDA oversight applies when an AI-enabled product is a medical device. Review is tied to the product’s intended use and technological characteristics, not to AI as an abstract category.
Free tools Windows power users keep installed
One-click scans. No signup required.
Rank #2
- Book: deep medicine: how artificial intelligence can make healthcare human again
- Language: english
- Binding: hardcover
The FDA describes a risk-based approach and uses different device pathways, including 510(k) clearance, De Novo classification, and premarket approval. These are distinct legal pathways; “FDA authorized” is a useful broad description, but it should not be substituted for the specific status of an individual device. The agency also reviews some modifications that could significantly affect a device’s safety or effectiveness.
The FDA’s public AI-enabled device list describes listed products as having met applicable premarket requirements, including a focused review of safety and effectiveness in relation to intended use and technological characteristics. The list is periodically updated. Being listed is a regulatory milestone for the specific device and use reviewed—not a blanket guarantee of safety, effectiveness, fairness, or benefit in every population or workflow.
How to read the current FDA count
The FDA’s overview reported more than 1,600 AI-enabled medical devices authorized for marketing in the United States as of September 2026. This is a dated count of marketing authorizations, not a count of devices shown to improve patient outcomes, products in routine use, or all AI applications in healthcare worldwide.
Rank #3
What kinds of devices make up the FDA-authorized landscape?
A 2025 peer-reviewed taxonomy by Singh and colleagues examined 1,016 FDA AI/ML device authorizations through December 20, 2024, representing 736 unique devices. In that study population, devices were classified both by clinical task and by the type of data their core algorithms used. The distinction helps explain why counts of authorized medical devices should not be generalized to all healthcare AI.
| Core algorithm input | Devices in the study | Share of 736 unique devices |
|---|---|---|
| Images | 621 | 84.4% |
| Signals | 107 | 14.5% |
| Omics | 5 | 0.7% |
| Tabular electronic health record data | 3 | 0.4% |
These figures describe the taxonomy’s review population through December 20, 2024, not the FDA’s later September 2026 total. They show an image-heavy profile among the studied FDA-authorized devices. They do not show that image-based AI dominates all healthcare software, AI research, or worldwide clinical practice.
Assessment and intervention are different tasks
The taxonomy grouped device functions broadly into patient assessment, such as diagnosis or monitoring, and intervention, such as surgery or treatment guidance. These categories describe what a device is intended to do; they do not establish that all devices within a category have the same level of autonomy, risk, or evidence.
How can you assess a specific healthcare AI application?
When evaluating a claim about an AI tool, identify the particular task and context rather than relying on a general claim that it “uses AI.” Four questions help keep comparisons meaningful:
- What clinical task is involved? Is the system assessing a patient—for example, supporting diagnosis or monitoring—or guiding an intervention such as treatment or surgery?
- What data does it use? The input might be an image, a physiological signal, omics data, or tabular health records. The FDA-authorized-device taxonomy is heavily weighted toward image inputs, but that does not make findings from one input type transferable to another.
- What is its intended use, and who uses it? Check the specific condition or function, intended user, and setting. A tool designed to assist a trained clinician in one context should not be assumed appropriate for another user or workflow.
- What evidence and regulatory status apply to that use? Distinguish the device’s specific FDA pathway and reviewed use from evidence of performance or patient benefit in a particular population and real-world setting.
These questions prevent two common errors: treating every AI product as if it were a regulated medical device, and treating authorization of one device as proof for a broader class of AI tools.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Why does authorization not settle whether a tool improves care?
Regulatory review and evidence about real-world effectiveness answer related but different questions. FDA authorization concerns a particular device, its intended use, its technological characteristics, and the applicable premarket requirements. Whether the device improves outcomes in a specific clinical workflow depends on evidence relevant to that use, population, and setting.
The FDA device count and the 2025 taxonomy are useful for understanding the authorized-device landscape, but neither is a comparative measure of patient outcomes, cost savings, clinician adoption, or performance across commercial products. Those claims require use-specific clinical studies or implementation evidence. A regulatory milestone should not be presented as proof of those broader effects.
What governance is needed as health AI changes?
Responsible use requires attention throughout a system’s lifecycle—not only when it is developed or first reviewed. The FDA emphasizes development, validation, deployment, monitoring, maintenance, and modification. A tool’s behavior and usefulness need to remain relevant to the context in which it is used, and significant changes may have regulatory implications.
WHO raises broader concerns about governance, ethical standards, regulation, and equitable access. Its 2023 publication, Regulatory considerations on artificial intelligence for health, is a stakeholder resource; WHO explicitly says it is not intended as guidance, a regulatory framework, or policy. WHO’s global governance discussion and U.S. FDA device authorization address distinct questions and should not be conflated.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Equity is also a practical concern: the benefits and risks of an AI application may not be shared evenly across settings or populations. A claim about a tool’s performance or value should therefore be tied to evidence for the people and circumstances in which it is intended to be used, rather than inferred from the technology label or authorization count.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
