Software based on artificial intelligence has been developed to improve the efficiency of medical treatment using electrified gas streams known as plasma. The computer code predicts the chemical substances emitted by plasma devices that can be used in cancer treatment, stimulation of healthy tissue growth, and surface sterilization.
The software has learned to predict the cocktail of chemical substances emitted by the stream based on data collected during real-life experiments and using the laws of physics as guidelines. This type of artificial intelligence, known as machine learning, involves the system learning from provided information. The researchers involved in the project published an article about their code in the Journal of Physics D: Applied Physics.
The plasma studied in the experiments is known as cold atmospheric plasma (CAP). When the CAP stream is activated, numerous chemical species in the plasma participate in thousands of reactions. These substances alter the treated cells in various ways, depending on the chemical composition of the stream. While scientists know that CAP can be used for killing cancer cells, wound healing, and bacteria eradication on food, the exact reasons why are not fully understood.
“This study is a step towards gaining a deeper understanding of how and why the CAP stream works, and in the future, it may also be used to improve its application,” said Yevgeny Raitses, the principal scientist at the Princeton Plasma Physics Laboratory (PPPL), a part of the US Department of Energy.
The project was carried out by the Princeton Collaborative Low Temperature Plasma Research Facility (PCRF), a collaboration between researchers from PPPL and George Washington University (GWU).
The software utilizes an approach known as physics-informed neural network (PINN). In PINN, data is organized into nodes and neurons, mimicking the way information is processed in the human brain. The code also incorporates the laws of physics.
“Understanding what comes out of the stream is very important. Knowing what comes out accurately is very challenging,” said Sophia Gershman, the lead research engineer at the PPPL’s PCRF, who worked on this project. This process would require several different devices to gather various types of information about the stream.
Calculating the chemical composition in the plasma is secured to nanoseconds
Li Lin, a research scientist from GWU and the lead author of the article, said that it is also difficult to calculate the chemical substances in the CAP stream because interactions have to be considered on a nanosecond level.
“When you consider the fact that the device operates for several minutes, the number of calculations makes the problem not only computationally intensive, but practically impossible,” Lin said. “Machine learning allows us to bypass the complicated part.”
The project began with a small set of real-life data collected using a technique called Fourier Transform Infrared Absorption Spectroscopy. The researchers used this small dataset to create a larger dataset. These data were then used to train a neural network using an evolutionary algorithm, which is a type of code inspired by nature that seeks the best answers through a “survival of the fittest” approach.
Ultimately, the team was able to accurately calculate the concentration of chemical substances, gas temperature, electron temperature, and electron concentration in the cold atmospheric plasma based on the data collected during real-life experiments.
In cold atmospheric plasma, electrons – small, negatively charged particles – can be very hot, while other molecules are close to room temperature. Electrons can have such low concentration that the plasma does not seem hot or burn the skin, while still significantly impacting target cells.
The possibility of personalized plasma treatment
Michael Keidar, a professor of engineering at George Washington University and frequent collaborator of PPPL who also worked on this project, said the long-term goal is to be able to perform these calculations fast enough for the software to automatically adapt the plasma during the procedure to optimize the treatment. Keidar is currently working on a prototype of such a “plasma adaptive” device in his laboratory.
“In an ideal case, it can be personalized. We imagine treating a patient, and each patient’s response will be different,” Keidar explained. “You can measure the response in real-time and then try to adjust the appropriate settings in the plasma-generating device based on feedback and machine learning.”
Further research is still needed to perfect such a device. For example, this study focused on the CAP stream for a certain period of time.
FAQ section based on the main topics and information presented in the article:
1. What is artificial intelligence-based software?
Artificial intelligence-based software is a computer program that uses artificial intelligence techniques, such as machine learning, to process data and make decisions.
2. How does artificial intelligence-based software improve medical treatment efficiency?
The software has learned to predict the chemical substances emitted by the plasma stream, which can be used in cancer treatment, stimulation of healthy tissue growth, and surface sterilization. This allows doctors to better customize procedures and optimize treatment.
3. What is cold atmospheric plasma (CAP)?
Cold atmospheric plasma (CAP) is a type of plasma that can be used for killing cancer cells, wound healing, and bacteria elimination.
4. How does the software learn to predict the chemical substances emitted by the CAP stream?
The software utilizes an approach known as physics-informed neural network (PINN). It leverages data collected during experiments and the laws of physics to learn how to predict the chemical composition of the CAP stream.
5. How does the software optimize plasma treatment?
The long-term goal of the software is to automatically adjust the plasma during the procedure to optimize the treatment. The software has the potential for personalized treatment as each patient’s reaction to the plasma can be different. The software can measure the patient’s response in real-time and adjust the settings of the plasma-generating device based on feedback and machine learning.
6. How far along is the project in developing the “plasma adaptive” technology?
The project on the “plasma adaptive” technology is still in the research phase. The researchers focused on the CAP stream for a certain period of time and further research is needed to refine the device.
Related links to the main domain:
– pppl.gov – Princeton Plasma Physics Laboratory
– gwu.edu – George Washington University
The source of the article is from the blog portaldoriograndense.com