New study from the Oregon State University College of Engineering shows that machine learning techniques may offer powerful new tools to advance personalized medicine, care that optimizes patient outcomes based on unique aspects of their care. biology and characteristics of their disease.
Research with machine learning, a branch of artificial intelligence in which computer systems use algorithms and statistical models to look for trends in data, tackles problems long unsolvable in biological systems at the level cellular, said Brian D. Wood of Oregon State, who conducted the study with OSU Ph.D. student Ehsan Taghizadeh and Helen M. Byrne of the University of Oxford.
“These systems tend to be very complex – first because of the large number of individual cells and second, because of the highly nonlinear way cells can behave,” said Wood, professor of environmental engineering. . “Nonlinear systems present a challenge for scaling methods, which is the primary means by which researchers can accurately model the larger-scale biological systems that are often the most relevant.”
A linear system in science or mathematics means that any change to the input of the system results in a proportional change to the output; a linear equation, for example, might describe a slope that gains 2 feet vertically for every foot of horizontal distance.
Nonlinear systems don’t work that way, and many systems in the world, including biological systems, are nonlinear.
The new research, funded in part by the US Department of Energy and published in the Journal of Computational Physics, is one of the earliest examples of the use of machine learning to solve the modeling problems of nonlinear systems and understanding the complex processes that might occur in human tissue, says Bois.
“The advent of machine learning has given us a new tool in our arsenal for solving problems that we couldn’t solve before,” he explained. “While the tools themselves are not necessarily new, the particular applications we have are very different. We are starting to apply machine learning in a more compelling way, which allows us to solve physical problems that we had no way to solve before. “
In modeling cell activity within an organ, it is not possible to individually model each cell of that organ – a cubic centimeter of tissue can contain a billion cells – the researchers therefore rely on this. called scaling.
Scaling seeks to reduce the data needed to analyze or model a particular biological process while maintaining the precision – the degree to which a model accurately reproduces something – of basic biology, chemistry, and physics. occurring at the cellular level.
Biological systems, Wood notes, resist traditional scaling techniques, and this is where machine learning methods come in.
By reducing the information load for a very complicated system at the cellular level, researchers can better analyze and model the impact or response of these cells with high fidelity without having to model each individual. Wood describes it as “simplifying a computational problem with tens of millions of data points by reducing it to thousands of data points.”
The new approach could pave the way for potential treatments for patients based on the results of a digital model. In this study, the researchers were able to use machine learning and develop a new method to solve classical nonlinear problems in biological and chemical systems.
“Our work capitalizes on what’s called deep neural networks to augment nonlinear processes found in transport and reactions within tissues,” said Wood.
Wood is collaborating on another research project using machine learning techniques to model blood flow in the body.
“The promises of individualized medicine are quickly becoming a reality,” he said. “The combination of several disciplines – such as molecular biology, applied mathematics and continuum mechanics – is combined in new ways to make this possible. One of the key elements of this will certainly be the continued advancements in machine learning methods. “