What do biology and data science have to do with one another? To Zhenyuan Lu, everything. As a research scientist at Northeastern working towards his PhD in industrial engineering with a focus on machine learning, Lu is applying the data processing and pattern-recognition power of AI to the fields of healthcare and disease diagnosis. Thanks to his background in bioengineering research and passion for data science, he feels uniquely qualified to tackle this challenge.
Lu, a Shanghai native and graduate of China’s Southwest University of Science and Technology, previously worked as a research scientist in healthcare and informatics before moving to the US to continue his education, where he has completed two masters’ degrees, an M.S. in Biology from Texas State and an M.S. from Northeastern in data analytics engineering. “It’s a good combination,” says Lu, “On the one hand is bioengineering, and on the other data science, and I found I could combine them together to do some good.”
That desire to do good arose from tragic circumstances in Lu’s family life. In the last decade, three members of Lu’s family have been diagnosed with cancer, the most recent an aunt who was diagnosed with terminal late-stage stomach cancer, a blow that occurred as Lu was completing his second masters and was considering multiple job offers. “That was one of the most important motivations for my PhD research,” he recalls, “I realized by working on disease diagnosis for serious diseases such as cancer, we may be able to detect it in an earlier stage. In my family’s case, it was detected only at a very late stage, too late for recovery.”
Rather than pursue a lucrative job, he turned down his offers and returned to Northeastern to begin work on his PhD, now intent on using his background in bioengineering and data science to someday save lives. Since 2019 he has done exactly that with Professor Sagar Kamarthi in Northeastern’s College of Engineering, where his work over the last two years has focused on applying machine learning techniques to the complex challenges of disease diagnostics and pain research. His current focus is on the latter area, specifically the question of quantifying how people experience pain.
“Pain research is quite important, because pain is one of the key data types doctors have to assess disease progression and treatment options,” Lu explains, “But all people experience pain differently, and for those with illnesses such as cancer, pain levels can be difficult to measure. We want to fix this by developing a standardized measurement system using machine learning and artificial intelligence.”
One of the primary challenges, as Lu describes it, is the huge variance in how people experience pain based on factors like gender, age, ethnicity or medical conditions. With so much variation in the data, until the advent of machine learning it would have been difficult or impossible to identify actionable patterns emerging from a diverse, global population’s responses to pain. Not so any longer; using a combination of current medical data collection methods, sensor data collected by tracking electrical signals in the patients’ brains and patient interviews, Lu and Professor Kamarthi are developing large libraries of signal processing data that are then fed into their AI model.
This model was first trained on signal data collected from research volunteers’ biological responses to stimuli such as contact with cold or hot water; using inputs such as blood pressure, skin temperature and respiration rates to establish their baselines, their model is now able to generate highly standardized datasets for them to study. “By combining the data from this bio-index with data from the doctors and the patients themselves,” Lu explains, “our AI model is now able to identify patterns.”
Armed with the data generated by their model, Lu and Professor Kamarthi are now developing a standardized scale that will be able to give doctors reliably accurate data on patients’ pain levels. The benefits to both doctors and their patients could be huge, says Lu. “In the future, patients could simply input bio-index information into a form and get a score that will inform their doctor’s treatment plan for them.”
When he completes his PhD, Lu plans to continue doing research in the pharmaceutical industry, where his work can continue to help people, either through his AI model or other work. His dream, he says, is to be able to work at the frontline of human health, where large-scale data science is applied to emergent illnesses such as COVID-19, where stakes are high and accurate predictions about disease behavior can save many lives. “I really admire and respect those scientists building predictive models at the front,” he says, “and I want to do that too. I want to know that my models can help to make big changes.”
Aside from his professional ambitions for his work, Lu is frank that his top priority has always been familial. “My first goal for my work is to help my family in the future. If some of my family have gotten cancer, it likely means that there is a gene affecting them, and other family members have the potential to get cancer as well. So if my predictive models can help them in the future, that is good.”