Grand Challenges Seek to Reset Bioengineering Agenda

Annual Meeting, BMES News,

This is three in a series of articles highlighting some of the technologies, processes and keynote plenary sessions presented at the 2024 Annual Meeting of the Biomedical Engineering Society, October 2024.   

While clicking across the Internet, youve probably come across headlines that include the phrase, This changes everything. They promise ground-breaking revelations or insights that will change your thinking. 

Thats exactly what the session, BMES Grand Challenges: Defining the Future of the Field, delivered at this years annual meeting. The two-hour panel discussions covered topics as diverse as personalized organ avatars, on-demand tissues and organs, reengineering the genome, and brain-interface systems.  

A decade or two ago, these may have sounded like science fiction. Today, they are realizable goals. Yet, as the BMES panels noted, there are substantial barriers we must hurdle to get there. 

The work grew out of a project begun during the pandemic by the IEEE Engineering in Medicine and Biology Society and the biomedical/bioengineering departments at Johns Hopkins University and University of California, San Diego. The initial team included biomechanical and biomedical engineers and also IEEE experts in AI, computation, and big data. They reached out to 50 noted researchers, then wove their input into five major challenge areas: 

Creating digital avatars of an individuals cells, tissues, and organs and using these models to predict health, wellness, and disease response. 

Making stem cell technology available to everyone so we can study how diseases affect individuals and to grow implantable organs that augment or replace their human counterparts. 

Developing exocortical technologies to measure and simulate brain function and interfaces for implants or devices to repair or augment the brain. 

Augmenting the immune system to improve health and wellness by quantifying immune response and engineering immunotherapies and next-generation vaccines. 

Engineering genomes and cells to cure diseases by building better ways to deliver genetic medications and control epigenetic outcomes. 

The BMES session consisted of two different panels. The first addressed biomechanical and biomedical engineering challenges while the second looked at issues related to AI, computation, and big data.  

Biomedical Barriers

The first roundtable started with a with a question to Gordana Vunjak-Novakovic, University Professor of Biomedical Engineer at Columbia University and this years BMES keynote speaker. Today, she explained, we are rapidly improving our ability to grow human tissues. By studying the growth process and how different types of cells, tissues, and organoids interact with one another, we can begin to amass the data needed for create digital avatars, said 

Yet there are major holes in our understanding. The first involves the immune system. It plays a role in absolutely everything, not just immune system diseases, she said. Her own work seeks to create bone tissue that produces immune cells. This enables her to study how the immune system interacts with samples of cancer or other diseases.  

The second gap involves blood vessels, a key way for organs and tissues in the body to communicate with one another. We are still very far from having functional vascularization of blood vessels that are perfused in the right way, she said. Without them, we can neither deliver the oxygen and nutrients needed to grow larger, more realistic organoids nor properly study organ-organ interactions. 

Brenda M. Ogle, Department Head of Biomedical Engineering at University of Minnesota, also had a wish list. The field has made progress genetically correcting mutations in single genes that cause diseases. Most genetic diseases, however, involve multiple genes.  

We want to know how we can deliver genetic medicines to more than one gene at a time and how much of the entire genome we can control, she said. Can we build new genes ourselves? Can we control transcription factors? As we amass all this data, how do we tap into it to help us answer those questions. 

She also noted that induced pluripotent stem cells are susceptible to genetic drift. Since they are critical bioengineering building blocks, it is important to learn how to keep them stabile over time. 

Probing the brain comes with its own set of challenges. Currently, there is a tradeoff between a devices invasiveness and its recording and stimulation resolution, said Anqi Zhang, who joins Caltech as an assistant professor of medical engineering in March.  

Some of her work focused on less invasive ways to implant high-fidelity monitors in the brain. Yet even the best implantable neurological interfaces have only thousands of channels. The brain has billions of neurons. We need to find ways to scale up brain interfaces to match the scale of the brain, she said.  

The brain also has thousands of distinct types of cells. We have no way to recognize those differences. Zhang faced that problem while trying to target a neuron associated with Parkinsons disease. She believes there may be a way to do this by modifying electrodes through genetic engineering. 

Finally, there is the brain data problem. Brain data is very noisy. If we could listen to millions or tens of millions of neurons, how would we sift this data? This is essential to study how human intelligence works and perhaps discern the first signs of neurological disease. 

An audience member asked about system complexity. While most research emphasizes how molecules communicate through the vascular system, human health is also governed by the lymphatic system, the microbiome, the nervous system and more. How do we measure that, he asked. 

This is a list of very, very difficult things to do, which explains why they havent been done, Vunjak-Novakovic said. Engineers have taken two different approaches to this problem, she continued. One is to print cells, tissues, and organoids and build from the bottom up. The other is to take a more holistic approach, tweaking cell circuits to drive them to form various systems. She believes that ultimately, both approaches will converge 

Computational challenges 

The second BMES session focused on computation, AI, and data.  

Researchers have access to far more data than ever, said Alexis Battle, professor of biomedical engineering and computer science at Johns Hopkins University. In computational genomics ten years ago, we were lucky to get data from 1,000 individuals, he said. Now we have data from hundreds of thousands and even millions of people. Its not only grown in quantity but in the data types that are available. Yet, for the vast number of the 50,000 genetic variations that separate you from the person sitting next to you, we have no idea what they do. 

Battle wants to understand how those variants respond collectively to our environment. We read our genomes differently at different stages of life and in different environments, so context really matters, she said. She believes AI software will help untangle how and when those genes are expressed, ultimately allowing physicians to predict what will help their patients. 

Not all data sets are compatible, often for reasons that are hard to remedy. Frederick Epstein, chair of University of Virginias Department of Biomedical Engineering, made that point while describing his experience with cardiac resynchronization therapy. This involves implanting a heart pacemaker. When it works, it has proven highly effective, yet 40 percent of people are unresponsive, he said.  

Epstein thought AI could help. His team collected data from 300 patients, taking vital signs and images for several years after the procedure. They then trained a machine learning system on this data to see if it could predict who would likely benefit from the procedure.  

The AI proved it could help physicians make better choices. Epstein thought the algorithm would prove even more accurate if it trained on data from thousands or tens of thousands of patients from other hospitals as well. Yet that data proved incompatible. The other hospitals did not conduct the procedure the same way Epsteins team did nor did they place the imaging coils on their patients chest in an identical manner. The algorithms predictive capacity also suffered when standards for drug treatments changed.  

Yet data and AI are just too powerful to ignore. They are here stay.  

The nature of any grand challenge is to emphasize what we dont know to encourage new procedures, new tools, new protocols, and new ways of thinking. While biomedicines grand challenges are imposing, thousands of researchers around the world are chipping away at them. They are finding ways to integrate these ideas into the way everyone does engineering and science.