Doctor of Philosophy, The Ohio State University, 2021, Computer Science and Engineering
Life sciences literature is replete with detailed and not-so-detailed instructions for wet-lab processes, called protocols, that communicate biological experiments to the scientific community. Nevertheless, due to the manual execution of these protocols, over 70% of researchers have failed to reproduce another scientist's experiments, with more than 50% unable to reproduce their research. An estimated $28B/year is spent on research that is not reproducible, with about 11% attributed to execution errors. Hence, there is a significant reproducibility and scalability crisis in scientific research. A researcher can spend weeks or even months setting up, optimizing, and validating new experimental techniques. And thus, he/she can at best realize the experiments in minimal ways (small sample sizes, etc.).
With an ever-increasing need for reproducibility and error-free replication of experimental procedures, laboratory automation is becoming increasingly crucial in many sectors of life science research. However, compared with manufacturing and service industries, the life science research industry is lagging in utilizing large-scale industrial automation for productivity, capacity, and quality improvements. Technological advancements (e.g., AI, modern software architectures and best practices, and sensing) can spur the development of intelligent automation systems for experimental procedures at higher precision and throughput that can also provide a significant reduction in human error. However, currently offered solutions have not seen widespread adoption.
One of the barriers in the intelligent automation of wet lab protocols is that the vast majority of them are written in natural language that effectively disseminates practical procedures within the research community but is difficult for automation systems to interpret. Through years of experience, life science researchers can naturally interpret wet lab instructions by understanding sentence structure, grounding (open full item for complete abstract)
Committee: Raghu Machiraju PhD (Advisor); Huan Sun PhD (Committee Member); Rachel Kopec PhD (Committee Member); Eric Fosler-Lussier PhD (Advisor)
Subjects: Artificial Intelligence; Computer Engineering; Computer Science; Experiments; Microbiology; Molecular Biology; Molecular Chemistry; Robotics; Robots