Plasma Diagnostics through Experimental Data-Driven Predictive Modeling of Plasma-Induced Changes in Diverse Targets

Low-temperature plasma (LTP) ignited at atmospheric pressure is emerging as an alternative radiation type that offers excellent capabilities in diverse applications ranging from medicine to material processing. Many of the physical aspects, such as reactive species formation, thermal properties of plasma, absorbed dose rate of plasma, diffusion properties of reactive plasma species, etc., are crucial for dictating the efficacy of plasma for numerous applications. The primary goal of my research is to develop a novel methodology to perform the diverse diagnostics of these aspects based on experimental data-generated predictive models of plasma-induced changes in diverse biologically relevant targets such as DNA and agarose gel.

The dissertation includes an overview of the various experimental and supervised machine-learning models I used to generate predictive models of plasma-induced changes in DNA and agarose gel targets. Then, I discuss the diverse diagnostics capabilities I achieved using predictive modeling of plasma-induced DNA damage, including plasma's gas temperature, reactive species generation, and dose rate of LTP. The qualitative diagnostics on the reactive species generation and the gas temperature were made from the predictive modeling of strand breaks and denaturation of DNA, respectively. The dose rate assessment capability I achieved using the DNA probe involved a creative strategy that combined the predictive modeling of DNA damage with correlations of absorbed dose versus DNA damage for other radiation types. The results indicated a remarkably high dose rate that can be further enhanced by tuning various process parameters, such as applied voltage and frequency, which would be beneficial for inducing radiobiological effects in a controllable manner.

In addition, my dissertation focuses on the colorimetric detection of the penetration of reactive oxygen species (ROS) in an agarose gel target, a tissue phantom. The predictive model of the penetration extent of ROS was used to derive specific metrics analogous to the ROS diffusion coefficients, and the impact of applied voltage and frequency on these metrics was investigated.

In this dissertation, I demonstrated the potential to perform diagnostics of crucial plasma properties from the predictive modeling of plasma-induced changes in targets. My study opens a potential gateway for developing novel plasma diagnostics tools that combine the power of predictive modeling tools such as machine learning and alternate probes like DNA.