Among the many applications of artificial intelligence (AI) in healthcare, devices have seen particularly strong momentum. AI-enabled medical devices — analysing scans, monitoring patients, and supporting clinical decisions — are moving quickly from development into day-to-day use.
More than 1,000 products have now been cleared by the US Food and Drug Administration — most of them in just the past five years. These tools, ranging from advanced imaging platforms to cardiac analysis software, promise faster diagnoses and more efficient care.
Yet alongside this rapid expansion, attention is turning to how these devices are tested and validated. Evidence of clinical effectiveness is increasingly seen as the factor that will separate technologies that earn lasting trust from those that struggle to gain traction. Recent research has underscored that devices without rigorous validation are more likely to run into problems once deployed — a reminder that proof of performance in real-world settings is just as important as the innovation itself.
Why clinical validation matters
As the number of devices grows, the spotlight is shifting from innovation to reliability. Clinical validation — testing that demonstrates a device’s accuracy, safety and real-world usefulness — is the key factor that determines whether these tools can truly be trusted in patient care.
A recent study published in JAMA Health Forum of nearly 950 FDA-cleared AI devices underscored the point. It found that devices without clinical validation were significantly more likely to be recalled, often in larger numbers. Nearly half of all recalls occurred within a year of clearance, showing the risks of bringing products to market before they are fully proven.
The authors concluded that prospective clinical testing should become the gold standard for AI devices, given their complexity and potential impact on patient safety. For clinicians, investors and regulators, the evidence base is quickly becoming the measure that separates durable innovations from short-lived experiments.
From promise to practice
Some companies are already embedding validation at the core of their strategies. One example is Australian medtech firm Artrya Ltd (ASX:AYA), whose Salix® Coronary Anatomy platform uses AI to analyse CT angiography scans for signs of coronary artery disease.
After receiving FDA clearance earlier this year, Artrya signed a five-year, US $600,000 agreement with Tanner Health System in Georgia, covering integration across five hospitals and 30 physician practices.
Read more: Artrya signs US$600,000 contract with Tanner Health, marks first US revenues
That agreement was the culmination of an extended validation and product enhancement program. Artrya worked closely with Tanner’s IT team to test the platform in a controlled environment, refining its performance before clinical deployment. Senior clinicians including Dr Shazib Khawaja, chair of the Tanner Heart and Vascular Centre, and Dr Ben Camp, Tanner’s chief medical officer, played key roles in demonstrating the platform’s clinical utility and supporting its adoption.
By embedding this level of validation and collaboration into its rollout, Artrya is showing how rigorous evidence can underpin both safety and commercial momentum — building confidence through proof in practice, rather than racing to market.
Building confidence in the AI era
AI-enabled devices are set to play a transformative role in medicine, but their success will depend on trust. The JAMA study highlights what can happen when clinical validation is treated as optional. Artrya’s trajectory demonstrates the opposite — that embedding evidence and validation from the outset can support both patient safety and sustainable commercial growth.
For the next wave of AI devices, validation will not be a box to tick after the fact, but the foundation on which lasting adoption is built.