Machine Learning for Plasma Sterilization Optimization

Over the past several decades, low temperature, atmospheric pressure plasmas have become increasingly used for various medical 1 and agricultural 2 applications. Low temperature plasmas often interact with cell and tissues through a combination of reactive species and charged particles; when applied directly to tissue, as in the case of dielectric barrier discharges, electric fields and ultraviolet radiation also contribute. Despite many studies over the past several decades, device optimization remains challenging because of the multiple parameters in device operation, including, but certainly not limited to, fill gas, applied voltage waveform, flow rates, and humidity. This complicated design space makes it difficult to optimize device design using simulations since it may not be known a priori how parameter variation may alter the relevant physics or chemistry and subsequent interaction with tissue or biological cells.