Resolve Magazine Fall 2023 >> Making Sense of Machine Learning >> Stories >> Decoding disease
“We don’t know what causes chronic fatigue syndrome,” says Xuanhong Cheng (pictured, below), a professor of bioengineering and materials science and engineering. “One hypothesis is that it could be caused by a virus, similar to Covid. After the main symptoms associated with the infection are gone, people still experience this long-lasting fatigue.”
It’s a difficult condition for doctors to diagnose, she says, because it’s unclear whether the symptoms are caused by a psychological or physiological issue. But previous research has shown that chronic fatigue syndrome (CFS) appears to affect the muscle tissue.
Cheng and Yu Zhang, an assistant professor of bioengineering and electrical and computer engineering, are part of a team using electrical sensors to measure the differences between cells from healthy and diseased muscle tissue. “These differences could be used as physical markers that confirm muscle damage, and therefore, a physiological cause of patients’ suffering. In other words, it’s not in their heads.”
Those measurements are enormously complex, with each one containing more than 200 features. So the team uses machine learning to identify the distinctive features associated with healthy tissue and with diseased tissue, and then separate them into groups.
“In the past, we were just manually picking features to define group A versus group B, and that’s a very subjective process,” Cheng says. “Machine learning gives us a more comprehensive view of what is important in the parameters that we’ve measured, and a more objective comparison of the different groups.”
Preliminary results have proven their model can, in fact, separate the groups of tissue. The next step is working with human patients, and the ultimate goal, she says, is to identify a spectrum of characteristics that doctors can use for diagnostic purposes for those with CFS and long Covid.
Main image: rohappy/Adobe Stock
