Biomedical engineering’s next test: proving value where patients actually live

Biomedical engineering has quietly become the scaffolding of modern medicine, and one of its leading voices now argues that the field must be judged not by what its devices can do in the laboratory but by what happens to patients at the bedside. Writing in Frontiers in Medical Engineering, Josué Sznitman of the University of British Columbia and the Technion–Israel Institute of Technology lays out a set of grand challenges for a new journal section devoted to research aimed squarely at clinical endpoints — survival, complication rates and recovery of function — rather than technical performance alone.

The argument arrives at a moment when engineered systems are inseparable from hospital care. In the intensive care unit, mechanical ventilation, extracorporeal membrane oxygenation and temporary mechanical circulatory support substitute, at least in part, for failing organs. The COVID-19 pandemic underscored how central these technologies had become: at its peak, mechanical ventilation was arguably the most widely used form of acute organ support worldwide, and the experience has since fed into updated guidelines on respiratory support in acute respiratory distress syndrome. Research on ECMO has grown steadily, with international guidelines now backing venovenous ECMO for severe respiratory failure, while pooled data from randomized trials inform the use of circulatory support in cardiogenic shock.

Yet Sznitman describes organ support as a double-edged sword by design. Mechanical ventilation keeps patients alive while it can also injure the very lungs it is meant to protect. Extracorporeal circuits expose blood to large foreign surfaces and non-physiological flows, leaving thrombosis and bleeding among the most frequent complications. These harms, he argues, are intimately tied to engineering — to how mechanical loads, flows and interfaces translate into tissue damage — and cannot be captured by device settings or dose-based arguments alone. Instead, they demand experimental and clinical studies that link support strategies to biological markers of injury in the patient.

Three engineering challenges frame the bedside agenda. First, the field must move from device settings to patient outcomes, identifying which mechanical quantities — strain, flow or shear — predict injury in a given patient. Second, extracorporeal and circulatory support devices must become safer and simpler to deploy so they can spread beyond a handful of expert centres. Third, as automation enters the ward, closed-loop systems for ventilation or anaesthesia must be evaluated against clinical endpoints rather than technical metrics; a recent multicentre trial found no gain in ventilator-free days with automated closed-loop ventilation compared with protocolized conventional ventilation. Across all three, Sznitman calls for measurements and models tested early in patients alongside clinical researchers.

Rehabilitation presents a different but equally pressing frontier. Stroke, spinal cord injury, limb loss and musculoskeletal conditions affect hundreds of millions of people, with musculoskeletal disorders alone accounting for the largest share of the global need for rehabilitation. For many of these patients, recovery of function — not survival — is the endpoint that matters most. Assistive biomechanical devices such as orthoses, braces and soft wearable exosuits are increasingly designed to support joints and muscles after injury or in chronic conditions. Wearable exoskeletons have drawn growing research interest, relying on sophisticated designs and control strategies to assist walking, while brain-computer interfaces are being explored to restore upper limb function after stroke. A Cochrane review of electromechanical-assisted gait training now provides pooled clinical evidence of its benefit, and bionic limbs with neural interfaces that restore sensation and more intuitive prosthetic control have advanced from laboratory demonstrations towards clinical use.

Despite that progress, a sizeable gap persists between what these devices achieve in the laboratory and what patients experience in daily life. Sznitman is candid that engineering alone will not close it: patient selection, therapy dosing and reimbursement all play a role. Still, he identifies central challenges — moving from short-term laboratory assessments to first studies in patients with functional endpoints that matter to them, adapting devices to the variability between patients and to changes within a single patient over the course of recovery, and bringing the technologies into clinics and homes, where cost, usability and long-term adherence determine impact. The rejection rate of upper-limb prostheses, he notes, has changed little despite recent technological innovations.

The operating room tells a story of rapid transformation. Robot-assisted minimally invasive surgery is now routine in many specialties, and the research community has accumulated a decade of experience on how robots may assist — and increasingly act alongside — the surgeon. The question has shifted from whether robots can perform surgical tasks to how much autonomy they should be given and under which conditions. The first surgical robots with some level of autonomy have already been cleared by regulators, though most remain at the lowest levels of autonomy. Artificial intelligence and haptic feedback promise to improve precision and safety, but evidence of their benefit for patients remains limited.

Cardiovascular interventions are moving in parallel away from open surgery and towards catheter-based procedures, with engineering contributing on two fronts. Robotic and magnetically steered catheters are opening access to vessels that remain out of reach for manual navigation, and artificial intelligence is increasingly embedded in their navigation. Transcatheter aortic valve replacement has become a standard of care for a growing population of patients, and patient-specific computational models of valve deployment are increasingly explored to anticipate complications such as paravalvular leakage before the procedure. But the learning curve of operators, the cost of such systems and uneven access across hospitals and countries all shape whether patients ultimately benefit. Robotic platforms record vast amounts of procedural data — instrument trajectories, forces, images — that could serve as objective quality measures, provided they are linked to patient outcomes. The major challenge, Sznitman writes, is moving from feasibility to clinical benefit: defining how levels of autonomy should be validated, regulated and accepted, demonstrating that robotic and image-guided procedures improve clinical outcomes rather than only procedural metrics, and validating patient-specific planning tools for structural heart interventions across patient populations rather than one patient at a time.

What ties these threads together, he argues, is the humble pilot study. Large randomized controlled trials remain the gold standard of clinical evidence, but they are costly, take many years and arrive late in the life of a medical technology, long after the key engineering choices have been made. Meanwhile, the step that matters most to engineers — the first use of a device or procedure in patients — remains comparatively underreported, often deemed too clinical for engineering journals and too preliminary for clinical ones. Sznitman champions small clinical pilot studies: first-in-human, feasibility and early-phase studies in which an engineering contribution is tested in patients for the first time. Their aim is not to establish efficacy but to ask whether a technology can be used safely, whether it behaves in patients as it did in the laboratory, and which endpoints and patient groups a subsequent trial should target.

Examples already exist. A bedside pilot study in twelve patients with acute respiratory distress syndrome showed that ventilator settings guided by electrical impedance tomography reduced the mechanical power delivered to the lungs. In rehabilitation, a first-in-human study followed a single patient over many months, in clinic and at home, as a brain-spine interface restored natural walking after spinal cord injury. In the catheterization laboratory, a first-in-human study tested whether a robotic system could safely assist a structural heart intervention through robotic-assisted transcatheter edge-to-edge repair. Each study was small, but each answered precisely the questions that determine whether a larger trial is worth pursuing. The new section, Sznitman concludes, will particularly welcome first-in-human and feasibility studies of novel technologies, pilot studies linking engineering parameters to early clinical endpoints, and reports connecting laboratory and clinical data obtained with the same technology, following established reporting guidelines for pilot and feasibility studies. The benefit of biomedical engineering for patients, he writes, will be decided where research meets the patient — and the technology has matured to the point where the question is no longer whether it works, but whether, for whom and under which conditions it improves clinical outcomes.

Subject of Research: Grand challenges in biomedical engineering research aimed at clinical endpoints across intensive care, rehabilitation and surgery

Article Title: Biomedical engineering for clinical endpoints: grand challenges where engineering meets the patient

Article References: Sznitman, J. (2026). Biomedical engineering for clinical endpoints: grand challenges where engineering meets the patient. Frontiers in Medical Engineering, 4, 2030893. https://doi.org/10.3389/fmede.2026.2030893

Image Credits: AI Generated

DOI: 10.3389/fmede.2026.2030893

Keywords: biomedical engineering, clinical endpoints, mechanical ventilation, ECMO, rehabilitation, exoskeletons, brain-computer interfaces, surgical robotics, transcatheter interventions, clinical pilot studies, first-in-human studies, patient outcomes