Mehdi Kiani (Brad Ziegler/UMBC)

Personalized and precise treatments will improve patients’ quality of life in a fast-approaching future driven by AI, wearable tech, and other innovations. UMBC student and faculty researchers led by Mehdi Kiani, a professor in the Department of Computer Science and Electrical Engineering, are at the frontiers of these changes.

They recently teamed up with colleagues at New York Institute of Technology and Pennsylvania State University to develop a system that combines state-of-the-art, millimeter-sized medical implants, computational models, and machine learning to better monitor and treat stomach disorders. A grant from the National Institutes of Health will fund the work through 2029.

The research offers the promise of improving individual medical treatment for gastric disorders such as gastroparesis, a chronic condition causing nausea and unexplained vomiting that affects more than 1.5 million people in the U.S. It also has broader implications for improving our general understanding of how the nervous system controls organs. 

“This type of research is vital, because it addresses fundamental gaps in how we monitor and treat complex organ functions,” says Kiani. “By integrating advanced sensing, modeling, and intelligent control, we can move beyond today’s limited approaches toward precise, adaptive therapies. These innovations have the potential to transform patient care not only for gastric disorders but across many areas of medicine.”

Shrinking medical implants

Kiani has extensive experience developing advanced, wireless medical implants. While at Penn State prior to joining UMBC, Kiani and his team developed a device that could harness energy from magnetic field and ultrasound sources simultaneously. The dual-powered feature is important, the researchers say, because it means the device can harness enough power to operate even as it is shrunk to millimeter-sized dimensions and implanted in a living body, where safety concerns limit the frequency of electromagnetic radiation that can be used to power and communicate with the device. 

Shrinking medical devices makes implanting them less invasive. It also means that many devices can be implanted across a wide area in the body, improving the ability to both monitor and treat disease. 

Kiani holds a medical device that could harness energy from magnetic field and ultrasound sources simultaneously. (Brad Ziegler/UMBC)

As part of the new research, Kiani and his colleagues envision a network of multiple tiny devices, called “gastric seeds,” implanted in the submucosal tissue of the stomach. The seeds will wirelessly monitor the electrical signals in the stomach that control its rhythmic contractions. They can also deliver electrical stimulation to correct misfiring signals. 

The seeds will be linked to a wearable band wrapped around the outside of the body, and will use the dual magnetic field and ultrasonic channels to both receive power and transmit and receive data. 

Building a virtual stomach

In addition to developing advanced implantable medical devices, the team will also build a virtual stomach to model the complex electrical and mechanical dynamics of a real stomach. This information, in turn, will help determine how best to use the gastric seeds to deliver treatment. 

The team will first construct an intricate and accurate model on a personal computer, and then use data from that model to train a machine learning model that can operate using the limited computing power of the wearable band. The machine learning model will efficiently interpret the sparse signals from the gastric seeds to determine optimal electrical stimulation treatments in real time. 

The team will test the integrated system on anesthetized rats toward the end of the project. 

Aydin Farajidavar, a professor of electrical and computer engineering at New York Institute of Technology and director of the Integrated Medical Systems Laboratory, and Farnaz Tehranchi, an assistant professor of engineering design and innovation at Penn State, will lead the computational organ model and machine learning model design elements of the project.

For the machine learning dimension of the work, the researchers will use computational models called physics-informed neural networks, which have attracted increasing attention for their ability to combine data-driven learning with fundamental physical laws. “When enhanced with human-like learning strategies, such as self-learning and adaptive optimization, these networks can evolve into significantly more powerful analytical tools,” Tehranchi says.

The advanced framework will provide deeper insights into stomach dynamics and disease progression, supporting more precise and personalized clinical interventions, she explains.

“What excites me most about this research is its truly multidisciplinary nature, bringing together expertise needed to tackle medical challenges no single field can solve alone,” Kiani says. “It’s also inspiring to work with talented students and help shape their careers as we develop technologies that can meaningfully advance patient care and improve quality of life.”

• More Information about Electrical, biomedical, and computer science researchers team up to develop a ‘cybergut’