Biomedicine needs an engineering overhaul

It’s not easy for surgeons to differentiate between cancerous and healthy cells, so when Vasilis Ntziachristos realized fluorescent agents can be used to visualize cancer, he saw an opportunity. “We thought this could be valuable for surgeons,” says Ntziachristos, director of Helmholtz Munich’s Institute of Biological and Medical Imaging. Fluorescent contrast agents, administered in vivo and highlighted by specialized lights before being visualized on a surgeon’s screen, indeed helped guide tumour resections. And the team soon identified another application.

In clinical trials, Ntziachristos and colleagues showed that when the agents are sprayed into the oesophagus, they enhance the detection of early cancerous lesions by more than 30%¹ . With early detection key to long-term survival, Ntziachristos estimates the technique could help save tens of thousands of lives a year and billions in health-care costs.

In biomedicine, solution-oriented thinking like this is rare. Although scientists generate vast quantities of knowledge, this doesn’t often translate into benefits for patients. What’s needed, many now argue, is a solution-centred model — with bioengineering at its heart.

Think about the aeroplane. After the physics was worked out, it took engineers to get the idea off the ground. Bioengineers can do the same for biology. “What we need are people who make it their career not to discover, but to find solutions,” says Ntziachristos.

To help catalyse this idea, in July 2022 Nature and Helmholtz Munich hosted a three-day conference in Munich, ‘Bioengineering Solutions for Biology and Medicine’, which featured pioneers from diverse realms such as AI and systems biology, as well as two Nobel laureates in Chemistry and one in Physics.

“We may need a transformation, if not a revolution, in how biomedical research is done,” says Matthias Tschöp, CEO of Helmholtz Munich. “We need to be faster and more effective in not only discovering, but validating, optimizing and safely delivering reliable solutions for serious health challenges.”

Engineering solutions in biomedicine

Speakers at the conference elaborated on three priority areas where an engineering-led approach can accelerate translation: observation, computation and validation.

New imaging and sensing methods offer the promise of not just observing early disease, but continuous monitoring. Subtle changes to the vasculature, for example, could be used to monitor the progression of diabetes, leading to better control of the disease. Combined with -omics techniques to assess genomic and metabolic profiles, Ntziachristos’s Munich team, which is funded by the European Union, aims to develop early detection models of disease, disrupting the medical paradigm of treating late-stage symptoms. Instead of treating disease after symptoms emerge, the transition to disease can be monitored and timely interventions prescribed. “We’re going at lightning speed now we have the ability to collect all this information cell by cell,” says biomedical engineer Shana Kelley of Northwestern University during another conference presentation.

In terms of validation, organs-on-chips have the potential to ease another bottleneck in the translation pipeline: drug discovery. Most drug trials are lengthy, expensive failures, and the animal studies underpinning them are often poor predictors of human response² . Memory stick-sized chips seeded with human cells can mimic organ-level responses, enabling researchers to model disease and test drugs. “This can flip the drug development model on its head,” says Don Ingber, founding director of the Wyss Institute for Biologically Inspired Engineering at Harvard. One study found that human liver-chips were eight times more accurate than animal models at predicting drug toxicity, a finding that could be worth around US$3 billion a year to the pharmaceutical industry³.

Computational tools, meanwhile, are helping make sense of the data that laboratories, trials and new technologies are generating. “The idea I’m putting forward is to discover the governing models of medicine,” says AI expert Mihaela van der Schaar, director of the University of Cambridge’s Centre for AI in Medicine. Machine learning is beginning to generate models that can predict the course and best treatment of an individual’s disease. In one of the largest AI studies to date, involving more than a million participants, a new machine learning system outperformed the model routinely used by the NHS to predict the trajectory and treatment of breast cancer⁴. The same framework has been adapted to provide treatment guidance for other conditions, such as cardiovascular disease and cystic fibrosis, and has also been used during the COVID-19 pandemic to predict the demand for intensive care beds and ventilators⁵ .

Building a translational culture

“The challenge now,” says van der Schaar, “is for people in AI and biology to talk to one another. Machine learners should realise this is a ‘problem space’ that requires them to invent new, cool tools.” Better communication between disciplines is certainly vital if researchers are to narrow the gap between basic research and life-changing solutions.

“We want to use AI in these complex biomedical settings to rationalize disease progression and guide clinical interventions,” adds Fabian Theis, conference co-organizer and head of Helmholtz_AI.

In Munich, three interdisciplinary institutes are being established. TranslaTUM combines engineering and oncology, the Helmholtz Pioneer Campus fuses engineering and metabolism, while the Centre for Bioengineering will focus on solutions-oriented technology development. These centres aim to promote novel tools for accelerating discovery, and the translation of knowledge into patient benefit, says Ntziachristos.

“People are the secret sauce,” says Stephen Quake, founding director of the CZ-Biohub in San Francisco. A diverse mix of people, allowed to think creatively and autonomously, challenge authority and take risks. “That’s where you break the paradigm and make new discoveries.”

Ingber agrees, and the Wyss Institute has an embedded structure to accelerate translation. Researchers are encouraged to report discoveries early, receive feedback from an intellectual property attorney, and talk to team members with industry experience. There are clearly defined measures of success, such as products in the pipeline, IP portfolios and start-up companies. There are timelines, milestones and validation projects, and as the project grows, business developers and venture capitalists are brought on board to help de-risk and finesse ideas.

It’s certainly paying off . The Wyss Institute generates a quarter of Harvard University’s annual IP. “We call them ‘collaboratories’,” Ingber says of the institute’s multidisciplinary groups. The model here, and at other institutions that embrace bioengineering, shows that translation doesn’t have to be an arduous hit-and-miss process.

By bringing a range of experts together, Helmholtz Munich showcased bioengineering’s role as the link between lab bench and hospital bed, accelerating the journey from basic discovery to clinical benefit.

“The moments where great minds from all the key disciplines come together for exchange and discussion can drive those transformations,” says Tschöp. “We hope this conference will represent one of those moments.”