Clockwise from top left: Rahul Sharma, Rohith Haridas, Vrushab Dharimane and Vineeth Challagali are working on their capstone project with digital health company PhenoMx, focusing on creating an analytical pipeline to predict Alzheimer's Disease.
MALVERN, Pa. — The transition from academia to industry can be difficult. Coursework might not easily translate to problems faced in the working world. Penn State graduate students Vineeth Challagali, Vrushab Dharimane, Rohith Haridas and Rahul Sharma aren’t worried about that, though, thanks to their capstone project and PhenoMx.
The four students are enrolled in the Master of Professional Studies in Data Analytics program at Penn State Great Valley, which culminates in an all-encompassing capstone course.
“The pedagogy of the course is that a team-based project experience requiring a project management mindset, technical deliverables, and minimal viable prototype will prepare students for careers in industry,” said Youakim Badr, associate professor of data analytics, who is teaching the capstone course. “It’s also supplemented with a series of invited speakers to fill some of the competency gaps often cited in the required skill sets of new data science and machine learning graduates.”
Haridas previously worked on a project with PhenoMx, a digital health company that uses medical imaging to address health care issues, and knew the company was interested in collaborating with an academic institution. So, he began discussing options with Badr.
Collaborating with Badr would be beneficial for PhenoMx, given their similar research projects, and working on a capstone project with students could expand that impact. The company could further its reach, expand its research, and help prepare students to enter the working world, all of which were important aspects to Girish Srinivasan, co-founder and chief technical officer.
“I’ve always felt that a lot of the research gets lost in the research world,” Srinivasan said. “For me, personally, one of the missions of the company was to do transformational work.”
After some discussion, PhenoMx proposed a capstone project that would focus on building an analytical pipeline to predict Alzheimer’s Disease. Challagali, Dharimane, Haridas, and Sharma had worked with healthcare data in previous courses, but the PhenoMx project involved neuroimaging data, something with which they weren’t familiar.
The group considered a few other ideas, but PhenoMx’s proposal stood out because it would be challenging, have a real-world impact, and provide first-hand exposure to working in industry.
“The impact we will create with this project motivates me every day,” Sharma said. “The research mindset required to accomplish the goal has made our team think out of the box, considering the state-of-the-art technologies. This gives us a perfect ending to our master’s program, with a minimal gap between industry and academia.”
The prevalence and severity of Alzheimer’s provided extra motivation for the students — according to the National Institute on Aging, Alzheimer’s is the sixth leading cause of death in the United States, and may be the third leading cause of death for older people.
Using data from functional magnetic resonance imaging (fMRI) that analyzes interactions between brain regions, the group’s capstone project aims to predict if a patient will develop Alzheimer’s based on a pipeline that identifies early cognitive decline. Challagali, Dharimane, Haridas, and Sharma are working on all aspects of the pipeline’s development, which is another new experience for the group.
“This is kind of a different experience compared to all the projects we have done until now,” Challagali said. “Now, all four of us are working with various types of things, so we need to learn everything. … It gives us a bigger scope to learn more and add onto our skills.”
PhenoMx divided the project into modules, providing the students with links to online courses and open source material to aid the development.
One of the first steps involved processing fMRI data sets. While the medical imaging presented an initial learning curve, the additional resources PhenoMx provided helped the students feel more comfortable. In fact, neither Srinivasan nor the students felt the lack of experience with imaging it impeded the group’s ability to work with the data.
“The scope of data science and data analytics is not limited to only certain domains,” Haridas said. “I would never imagine that we can analyze brain activity and apply deep learning models to predict whether a person has a particular disease. It’s fascinating and understanding how the brain functions boosted our enthusiasm.”
Enthusiasm is something Srinivasan noticed, too. Because PhenoMx is a small company, employees work closely with the student on all aspects of the project. The group has a Slack channel and other methods of communication so employees can quickly answer questions the students have, avoiding long developmental delays.
“I see a lot of motivation and I’m encouraged that these students are quite diligent in going about doing their work, coming back, and telling us what challenges they face,” Srinivasan said. “I’m almost seeing them as our extended team for this project. As much as I’m interested in trying to make sure they understand and get the most out of it, I’m also relying on them, in some sorts, to delve into a few things.”
Srinivasan plans to work with Penn State Great Valley on more capstone and research projects and hopes to expand the PhenoMx’s interactions to other departments across the University. The collaboration provides more than project deliverables, though. As PhenoMx grows, Srinivasan expects the company will look to hire students who are already familiar with the intricacies of the projects and comfortable with the team dynamic.
Regardless of where Challagali, Dharimane, Haridas and Sharma begin working after graduation, they’re grateful for the opportunities the PhenoMx capstone project has presented them. Working with neuroimaging exposed them to new facets of the industry and, combined with the experience that comes from being immersed in a real-world problem, they feel confident as they prepare to enter the full-time workforce.
“This is something close to a real-world problem, something that happens in a company day-to-day,” Dharimane said. “I think this project is helping us to get to know what really happens in a company and how we have to be prepared for that. … For any industry, we need domain knowledge and this project is certainly giving us that for the healthcare industry.”
As data-driven roles and companies continue to expand, so, too, does the need for leaders with project management skills and multidisciplinary backgrounds in data science and machine learning. Long-term, early-stage research partnerships between academia and industry can help prepare students for the working world while also imparting important non-technical skills.
“The ability to work in teams, acquire skills to resolve conflicts in project management and evaluate ethical and fairness issues will be extremely helpful to our students to develop their soft skills in the era of data analytics,” Badr said. “At the end of this course, I am confident that our students will be capable of leading their own machine learning systems in production and understand the social and economic impacts of their systems on society as a whole.”
