WSU projects draw on AI to advance medical science

This story is part of an AI series looking at how WSU is driving innovation in research and teaching through artificial intelligence. View the entire series as it becomes available.

In the not-too-distant future, a web-based tool designed at Washington State University will put massive amounts of cancer data at the fingertips of researchers, public health officials, and clinicians - accelerating critical research into the disease that costs millions of lives every year.

Built on AI technologies, the informatics tool will enable researchers to analyze combined national datasets across a range of cancer-related characteristics. It will allow users to examine environmental and behavioral factors, compare molecular data and tumor characteristics across different cancers, and map cases at the state and county levels.

The project builds on previous research from the lab of WSU computer scientist Assefaw Gebremedhin, and it highlights the way AI innovations are expanding scientific knowledge at an exponential pace - innovations fueling further innovations.

"The progress around AI has been extraordinarily fast," said Gebremedhin, Berry Family Distinguished Associate Professor in the School of Electrical Engineering and Computer Science and head of the Scalable Algorithms for Data Science Lab. "It is almost difficult to even keep track."

Health and medicine have long been research priorities at WSU. AI is ramping up that work dramatically, in genomics, pharmaceuticals, immunology, smart health systems, and more - accelerating the analysis of ever-more data, and fueling change in how human health is studied and medical care is delivered.

These uses are widespread across many fields. Researchers in the College of Veterinary Medicine's Paul G. Allen School for Global Health have developed a machine learning model that identifies animals that harbor viruses that could spread to humans - which could help limit, or even prevent, future pandemics.

Computer scientists in the Voiland College of Engineering and Architecture have designed a self-improving AI model that improves 3-D printingthat could be applied to the creation of artificial organs, among other uses.

A grad student in molecular biosciences - frustrated at the tedium of manually handling strands of hair to study under a microscope - helped develop an AI tool that analyzes hundreds of hair samples in seconds, laying the groundwork for the future use of hair to diagnose human health.

One key frontier is the field of diagnostics, where AI is expected to have a major impact in improving accuracy and lowering costs. A deep learning model developed at WSU was shown to be much faster, and often more accurate, than humans in identifying signs of disease in images of animal and human tissue.

The model not only correctly identified pathologies but did so faster than previous models - and in some cases found instances that a trained human team had missed. It offers the promise of improved medical diagnosis, such as spotting cancer in a biopsy image in just minutes, a process that typically takes a human pathologist several hours.

"This AI-based deep learning program was very, very accurate at looking at these tissues," said Michael Skinner, a professor in the School of Biological Sciences with a long record of accomplishment in reproductive biology and epigenetics. "It could revolutionize this type of medicine for both animals and humans, essentially better facilitating these kinds of analysis."

Accelerating science

A key benefit of AI in the lab is speed - it allows the creation of tools that can crunch much more data, much more quickly, than traditional data-management methods.

A recent interdisciplinary example: Researchers studying viruses from the School of Mechanical and Materials Engineering and the Department of Veterinary Microbiology and Pathology wanted to examine thousands of molecular interactions involved when disease-causing viruses invade cells.

Undertaking a traditional series of experiments would have been tremendously time-consuming. The WSU team used an AI model and molecular modeling to more quickly identify and block a crucial molecular interaction that allows a herpes virus to enter cells, suggesting future possibilities for treating viral illnesses.

"It was just a single interaction from thousands of interactions," said Jin Liu, a professor in the School of Mechanical and Materials Engineering. "If we don't do the simulation and instead did this work by trial and error, it could have taken years to find."

The realm of AI in medicine also comes with a range of ethical considerations. The use of individual information in data sets raises privacy issues. Considerations about social inequities are heightened, given that AI tools are based on information fed into them - raising the prospect of exacerbating existing inequities. And fundamental questions remain about maintaining patient autonomy and control over their treatment as care becomes more automated, highlighting the continuing need for human engagement in treatment decisions.

Thomas May, the Floyd and Judy Rogers Endowed Professor and medical ethics program director in the Elson S. Floyd College of Medicine, discussed these challenges in a WSU web presentation in 2024. He highlighted an example of a patient who had suffered a stroke, whose doctors were recommending an elective surgery that would extend her years with her grandchildren but limit her physical activity.

But the patient prized gardening more than time with her grandchildren. Good medical care - with or without AI - would call for understanding and respecting her autonomy in making that decision, he said.

"A year or two of working in the garden was valued more than five or even 10 years of caring for her grandchildren - to her," May said in the presentation. "It's that sort of emotional element that is unlikely to be reflected in AI data, which is much more likely to look at the number of life years that could be expected."

The web-based tool developed in Gebremedhin's lab - now in the final stages of preparation for journal submission - expands upon previous work designing a system for analyzing the molecular drivers of cancer, making it easier to rapidly analyze vast troves of tumor data.

Now the benefits of that technology will be made widely available in the global fight against cancer.

"I am very excited about this work because population-level geospatial analysis of cancer incidence plays a critical role in identifying regional disparities and well-informed public health decision-making," Gebremedhin said. "Mapping cancer burden at state and county levels enables researchers and policymakers to detect spatial clustering, assess associations with environmental and lifestyle factors, and evaluate inequities related to socioeconomic conditions and healthcare access."

Tina Hilding, the director of communications in the Voiland School of Engineering and Architecture, contributed to this report.