As companies such as Elon Musk’s Neuralink begin human trials of high-risk brain implants, a new proposal calls for a major change in how the U.S. handles injuries caused by the devices. A “no-fault” compensation program could help patients harmed by devices like brain-computer interfaces (BCIs)—even when no one is legally at fault. These devices, implanted in the brain to treat serious conditions like epilepsy or paralysis, can offer life-changing benefits. But they come with serious risks such as seizures, strokes or even death. When something goes wrong, patients often have no way to get help. This proposal would provide a way to balance innovation with justice, and allow companies to push boundaries without leaving behind the patients who take these risks.

Zachary Sexton and colleagues have developed a design platform that can rapidly generate vasculature trees that can then be bioprinted and used to successfully perfuse living tissue constructs. The platform improves the design and production of complex vascular networks that will be needed for manufacturing human tissues and organs in the future. As the researchers note, the manufacture of tissues with multiple cell types has improved recently. But like a city needs a full complement of main highways, side streets, and alleyways to carry traffic its furthest reaches, complex geometric tissues and organs need a vasculature network that will reach and adequately perfuse blood throughout the tissue – a design feat that has remained challenging for researchers. Sexton et al. leveraged algorithmic advances to create a synthetic toolkit that can generate organ-spanning tree networks within minutes (a 230-fold acceleration over current platform speeds), with the ability to simulate flow patterns and pressures. The researchers created computational vasculature networks for more than 200 engineered and anatomical models. In a bioreactor, the researchers also tested the perfusion of 3D bioprinted networks, which improved the cell viability of the printed tissue constructs.

Although there is no direct case law on liability using medical AI, the recent products liability case Dickson v. Dexcom Inc. may hold some lessons on future liability risk for manufacturers incorporating AI or machine learning (ML) technologies, according to Sara Gerke and David Simon. This legal finding could influence which products in this space are developed and marketed, with impacts on research and development and resulting benefits to the public. In this Policy Forum, Gerke and Simon discuss the case, which is the first to hold that federal law can preempt personal injury lawsuits under state law when the Food and Drug Administration codifies regulations applicable to a Class II medical device authorized through its De Novo pathway. De Novo classification requests are made for devices that cannot seek the FDA’s Premarket Notification or 510(k) approval pathway – because there are no devices substantially equivalent to it on the market – but are not deemed risky enough to require FDA Premarket Approval (PMA). New technologies such as digital therapeutics are often approved under the De Novo pathway, and Gerke et al. note that “AI/ML-enabled device manufacturers are utilizing the De Novo pathways over three times more than most device manufacturers.” Dickson may shed light on how courts address liability issues with these devices in the future, the authors conclude.

By creating a cohesive AI-enabled and physics-based molecular modeling and drug discovery pipeline built around film industry animation software, researchers at the Wyss Institute at Harvard University identified the orally available, FDA-approved drug bemcentinib, as a potential antiviral agent. They then used this chemical compound as a launch pad for developing a more specific and effective antiviral drug with efficacy against a broad range of coronaviruses, which has great potential for fighting future respiratory pandemics.