Students, Professor Use Machine Learning to Explore Proposed Saliva-Based COVID-19 Test

Friday, August 28, 2020
Professor Michael Jo and a student pose next to computers while they are involved in a tele-meeting with other researchers.

Electrical and computer engineering assistant professor Michael Jo and students had a recent online meeting with 12-15 Molecular Diagnostics Chief Technology Officer Ewa Kirkor to review aspects of their research project.

An electrical and computer engineering professor and three students are applying elements of machine learning, artificial intelligence, and computational modeling to examine ways nanomaterials could improve the accuracy and response time of a promising saliva-based COVID-19 screening procedure created by Connecticut-based scientists.

If successful, 12-15 Molecular Diagnostics’ Veralize novel testing procedure using a carbon-based direct sensor for the first time, would be safer and less costly than current methods and easily disposable. Currently, still in development and clinical testing, the device provides coronavirus test results in approximately 20 minutes.

Assistant Professor Michael Jo used data-driven modeling techniques in doctoral studies to confirm the thermal properties within a specifically balanced composition of graphene-derived materials. Concurrently, this material helped found 12-15 Molecular Diagnostics by Chief Executive Officer Saion Sinha and Chief Technology Officer Ewa Kirkor.

Juniors Daniel Su and Xingheng Lin spent this summer working with Sinha and Kirkor to improve the machine learning framework within bionanosensor modeling aspects of the project to improve test response recognition. Senior Hailey Heidecker is joining the undergraduate research project this fall.

“Our team has joined at a very important time in this project,” says Jo. “We are working on constructing the first principle by simulation and to improve their pattern recognition of their test response.”

In hopes of gaining Federal Drug Administration certification, 12-15 Molecular Diagnostics is currently doing real patient testing against SARS-COV-2, the causative agent of COVID-19, and with further improvements, detecting a variety of influenza strains. A recent online product demonstration for Rose-Hulman faculty, staff, and students from a laboratory at East Haven, Connecticut confirmed that saliva testing could determine a result quickly and accurately.

“The students' modeling and machine learning work, under the guidance of Dr. Jo and with our collaboration, will help us achieve a better understanding of the sensing characteristics within our project,” says Kirkor. “It will also help with the formulation of requirements, design, and building of the prototype for prompt, direct, and simultaneous detection and differentiation among multiple organisms.”

She continued, “The students have broad interests in artificial intelligence and computational modeling and are ready to tackle structural features of the sensing assembly of nanocarbons and model biological materials that we are planning to use in our work. Moreover, the students are interested in making detection accurate and reliable through machine learning, features that we have included in Veralize, which can serve as a benchmark for testing and improvements of models and computational programs.”

Su, an electrical engineering student, has already proposed a modified resistive network model that has provided a better understanding of how the novel bionanosensor uses carbon nanotubes and graphites to detect COVID-19.

“Since we are at an age where technological advancements are so small and complex that a normal person can't calculate or predict the outcome of their machinery, we use a computer to do the heavy lifting and crunch all the numbers and tell us what we could do to maximize our process and make everything more efficient,” the junior says. “I'm currently working on the computer interface where we could plug in some numbers that are known and it would optimize our process and tell us what changes we need to make to achieve this goal.”

Su continued by stating, “I really like the creative side of research. I want to make an impact on COVID-19 and help as many people as possible. At this point, it is clear that we can't force everyone to comply with safety regulations, so I personally think that the best peaceful solution to this would be to increase testing. A positive COVID result is a stronger verification than a research study since we can't assume everyone understands science.”

Meanwhile, Lin will use machine learning methods to increase the project’s limited sample data size to improve detection accuracy. He also has worked with Professor Jo this summer on the development of a framework that can reduce image data size while, in turn, accelerating the speed of conventional machine learning and deep learning algorithms. A publication may result during the 2020-21 academic year.

“Health care and medical diagnostics are among the major applications of machine learning. I hope my knowledge and work can help control this global pandemic,” he says. “Getting the opportunity to work with Dr. Jo on research in this field, by getting a deeper understanding of machine learning, is a great opportunity.”

Heidecker, an electrical engineering major who has an interest in medical device research, will perform nano-scale simulation for carbon-based materials and cooperate with Su on the modeling the bionanosensor.

“When Dr. Jo told me of the work he was doing with bionanosensor modeling for COVID-19, I was instantly hooked,” she remarks. “Upon graduation, I plan on pursuing a doctorate in neural engineering. I believe this project will help me cultivate my research skills, while also introducing me to applications of electrical engineering within the medical field.”