Following Nobel Prize-winning chemist David Baker’s recipe for cooking an antidote to cobra venom using artificial intelligence (AI) could be faster and more effective than currently available methods. The ingredients and steps can be found in a new study published by the University of Washington (UW) scientist in collaboration with the Technical University of Denmark. They are ready for the next steps in preclinical trials.

Artificial intelligence (AI)-based models developed by a team of international researchers were able to identify ovarian cancer in ultrasound images more accurately than humans. Results from a study published in Nature Medicine showed that the AI models achieved an accuracy rate of 86.3%, compared to 82.6% for the experts and 77.7% for the non-expert examiners.

Some people may have seen 2024 as the year of artificial intelligence (AI) in med tech, but the FDA is off to a strong start in 2025 with a dual-purpose AI draft guidance. While the draft covers both premarket submissions and life cycle management considerations, the more important consideration is that the FDA’s centers for devices, drugs and biologics have all signed off on the draft, suggesting an agency-wide convergence in thinking about AI.

Regulatory overhauls are never a simple affair, but the med tech industry ran into an amorphous, ennui-inducing mess in dealing with the EU Medical Device Regulation almost immediately. Even though the underlying legislation passed in 2017, the MDR impasse continued to impede innovation on the continent – a problem that might be only partly resolved in 2025.

Taipei, Taiwan- and Watertown, Mass.-based Syncell recently raised $15 million in series A financing to advance its subcellular protein purification and spatial proteomics analysis product called Microscoop Mint, based on its Microscoop platform technology.

Heartbeam Inc. received U.S. FDA 510(K) clearance for its portable, non-invasive electrocardiogram system that enables on-the-go recording of heart signals. The cable-free, credit card-sized device uses five electrodes to capture signals from three directions and can be used whenever a patient feels symptoms, overcoming the challenge of identifying intermittent arrhythmias, which are often not experienced during scheduled appointments, without requiring patients to wear a device continuously.

Bring your dog to work day looks a little different at Spotitearly. The Tel Aviv-based company asks its canine workers to bring their whole selves to the job – particularly their finely trained noses – to help suss out cancer in breath samples. The dogs’ unique capabilities are interpreted by AI to capture canine indicators humans might miss and the combo has remarkably high accuracy, 93% or better in three of the four common cancers tested.

The health care potential for generative artificial intelligence comes with hazards such as the inadvertent effects of poor prompt engineering practices, but Anil Bhatta of Deloitte Consulting told a U.S. FDA advisory committee that this risk could be managed by rules that would thwart any consequent problems, such as an inadvertent jailbreak of the algorithm’s intended use.

Diagnosing depression and other mental illnesses is largely subjective, but that could soon change as Sydney-based Trivarx Ltd. developed an AI-based algorithm that uses heart rate and heart rate variability coupled with sleep disturbance to diagnose major depressive episodes.

Fifty years after the term brain-computer interface (BCI) was coined, Neuralink and several other companies are reviving the promise of BCI for health care, as artificial intelligence and novel methods of neural signal detection work hand-in-hand to propel biomedical innovation.