Building a data-driven approach with plant science | News

Machine Learning


For many people, photosynthesis is a fond memory from middle school or high school.

But in Northwestern Engineering’s new data-driven plant science course, photosynthesis is a live signal that can be measured, analyzed, and modeled using computational tools.

This course introduces students to data-driven approaches in plant science that bridge experimental biology with embedded sensing, bioinformatics, and machine learning.

Students like Miller-Watson, who started coding at the age of 9, were drawn to interdisciplinary courses as a new detour from their usual studies in computer science, computer engineering, and engineering design.

“The real strength of computer science is the ability to combine it with other fields and enhance other fields,” said Watson, a fourth-year computer science student. “And this class is a perfect example of that. Just a little bit of computer science knowledge can change the way you care for your plants. And thinking about intersections like this opens up so many new possibilities.”

Nivedita Arora, Kitten Lee, Susan StricklerThe course was launched this winter in partnership with the Joint Program in Plant Biology and Conservation at Northwestern University and the Weinberg College of Arts and Sciences at the Chicago Botanical Garden. Professors Nivedita Arora and Susan Strickler developed this course in collaboration with Qitong Li (MS ’25), a research specialist in Arora’s VAK Embodied System Lab. and course peer mentor Raj Dave, a master’s student in computer science.

Through lectures and hands-on labs, students in the Data-Driven Plant Science course learn to design experiments, collect and visualize sensor data (e.g., carbon dioxide, temperature, humidity) using Arduino microcontrollers, extract and process RNA, and apply computational techniques. This curriculum covers RNA sequencing and Linux-based bioinformatics, along with machine learning techniques for interpreting complex biological datasets.

“In this age of AI, one of the most important skills students need to learn through their courses and research activities is to engage in creative systems thinking across disciplinary boundaries,” said Arora, the Allen K. and Johnny Cordell Breed Junior Professor of Design and assistant professor of computer science and electrical and computer engineering in the McCormick School of Engineering. “And that’s what we’re aiming for.”

Rather than being given a dataset to work on, Arora and Strickler allow students to collect and understand data about the specific plant science question they choose to investigate.

“When I was studying plant biology in graduate school, I wasn’t thinking about sensors or how you could use electrical engineering and computer science in your experiments,” says Strickler, adjunct associate professor of plant biology and conservation and associate scientist in conservation genomics at the Chicago Botanic Garden. “I love this new world of approaches and classes that allow students to explore how different disciplines can work together.”

plant seeds

The origins of the course germinated two years ago, when Arora and her lab began thinking about how plants could act as living batteries for IoT devices. Arora, Strickler, and Lee, then master’s students in computer engineering, were particularly interested in plants that adapted to arid environments through crassulacean acid metabolism (CAM) photosynthesis. Succulents, cacti, orchids, and other CAM plants reduce water loss by opening their stomata (pores) at night to fix carbon dioxide (CO₂).

Their two-year collaboration, supported in part by the American Society of Plant Biologists and the Northwestern Institute for Complex Systems (NICO), led to the development of a low-cost (approximately $30) microcontroller-based toolkit designed for scientists, students, hobbyists, and farmers to investigate plant phenotypes.

“This is a low-power, frugal, accessible, and democratizable device for collecting real-time data,” Arora said. “Think of large, very expensive medical devices and relatively humble diagnostics with smartwatches. That’s what we’re trying to apply to non-destructive data collection in plants.”

Along with Sarah Jones (Chicago Botanical Garden), Lee, Arora, and Strickler pilot-tested the toolkit through a workshop they co-hosted last summer for the Chicago Botanic Garden’s National Science Foundation Research Experiences for Undergraduates (REU) Genes to Ecosystems internship program. Lee explained that this experience influenced subsequent course development.

The Frugal Plant Toolkit is in action

At a showcase event on March 12, student teams presented their final projects that integrated microcontroller toolkits with hardware, biological, computational, and DIY custom components. The five teams investigated topics such as heat wave resilience, biohybrid robotic systems, the genetics of photosynthetic modes, monitoring the health of houseplants, and sound-induced stress responses.

Plants as sensors in biohybrid robotic systems

The research team asked the question, “What if we could integrate the functionality of organic life forms directly into robotic systems?” For their research, they prototyped a robotic system that uses CO₂ as an indicator of plant health and environmental suitability to help move plants to optimal conditions. The team iteratively designed the motor-controlled breathing mechanism using CAD software, 3D printers, and laser cutters. This causes an iris door at the bottom of the plant’s closed vessel to open and close, allowing the CO2 to escape, allowing the gas to be measured at different locations in the room as a function of light proximity.

“Taking this course allowed me to explore my interest in green robotics and how living systems can create robots that are low-power, biodegradable, and have a low environmental impact,” Petrie said.



Source link