Over a million developers have joined DZone.
{{announcement.body}}
{{announcement.title}}

How AI Developed for Video Games Can Treat Sepsis

DZone 's Guide to

How AI Developed for Video Games Can Treat Sepsis

Researchers have used deep learning approaches commonly seen in video games to try and discover new therapeutic drug strategies for sepsis.

· AI Zone ·
Free Resource

I've written numerous times about the dangers posed by sepsis and the various technological solutions to the condition. One of the more interesting approaches has been taken by researchers at the University of Vermont, who have used the kind of deep learning approaches commonly seen in video games to try and discover new therapeutic drug strategies for sepsis.

The researchers have developed a system that treats the immune system simulation as a kind of video game, with outputs from the simulation forming the score that allows the neural network to manipulate 12 cytokine mediators to stimulate a response from the immune system and bring the infection down to normal levels again.

"It's a complex system," the researchers explain. "Previous investigations have thus far been based on manipulating a single mediator/cytokine, generally administered with either a single dose or over a very short course. We believe our approach has great potential because it explores much more complex, out-of-the-box therapeutic strategies that treat each patient differently based on the patient's measurements over time."

This allows the system to propose treatment strategies that are both personalized and adaptive as it evolves according to each individual patient. Each run of the simulation presents the system with a different patient type, who presents with different initial conditions.

A Smarter Approach

By using this deep reinforcement learning approach, the researchers were able to identify a new treatment for patients that achieved a 100 percent survival rate (for patients the system had been trained on at least), with a lower than 1 percent mortality rate for 500 patients selected at random.

"The simulation is mechanistic in nature, which means we can virtually experiment with drugs and drug combinations that haven't been tested before to see if they might be promising," the authors explain. "The number of possible treatment strategies is huge, especially when considering multi-drug strategies that vary over time. Without using simulation, there's no way to evaluate all of them. The hard part is discovering a strategy that works for all patient types. Everyone's infection is different, and everyone's body is different."

The team is so confident in their work that they believe it can also be deployed on various other conditions as well. They ultimately want to create a tool that allows a range of patient data to be fed into it, and treatment plans then produced that administer the right drug, at the right time and in the right dose.

Suffice to say, they're a long way from that at the moment, and a lot more testing needs to be undertaken, but the team are confident they'll get there.

"This is an exciting project," the team concludes. "This is an incredibly novel project that brings together three cutting-edge areas of computational research: high-resolution multi-scale simulations of biological processes, extension of deep reinforcement learning to biomedical research and the use of high-performance computing to bring it all together."

Topics:
artificial intelligence ,machine learning ,ai in healthcare ,technological solutions to sepsis ,deep reinforcement learning ,ai news

Published at DZone with permission of

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}