Surrounded by medical notes, senior Vrishank Chandrasekhar immerses himself in lines of code — developing an algorithm to help doctors create targeted cancer treatment plans. By analyzing clinical free text, or typed notes from medical practitioners, Chandrasekhar designed a digital biomarker for early pan-cancer survival and recurrence prediction. He was recognized as one of the top 300 Scholars in the 2025 Regeneron Science Talent Search, earning both himself and Lynbrook $2,000 awards.
“The prospects of using machine learning to revolutionize healthcare systems enraptured me,” Chandrasekhar said. “Seeing a gap in current prognostics through an intensive literature review further inspired me.”
A year before designing his digital biomarker, Chandrasekhar competed in the Regeneron International Science and Engineering Fair with a project in computational pathology. Applying data science to analyze images of cancerous tissue for patient biopsies, he developed an algorithm that can predict gastrointestinal cancer occurrence events. The feedback that Chandrasekhar received at the contest was invaluable, leading him to his latest project.
“I wanted to expand the scope of my prognostic model to predict the survival outcomes of patients across multiple cancers,” Chandrasekhar said. “I was also inspired to design a predictive biomarker with non-invasive data.”
Chandrasekhar dove head first into implementing his innovative goals. He switched from analyzing full-sized medical images to text-based clinical notes, challenging himself with a new approach. By applying advanced language processing algorithms to free-text data such as medical notes, Chandrasekhar identified digital biomarkers that could predict a patient’s likelihood of survival and risk of recurrence across multiple types of cancer.
“As an independent researcher, one of the biggest challenges I faced was a lack of mentorship specific to my field that I could really build off of,” Chandrasekhar said.
Without direct mentorship, tackling a niche problem like predictive health analytics was challenging. Realizing that he needed to build a solid foundation of the scientific concepts behind his project, Chandrasekhar made use of online resources such as seminar presentations, scientific literature, informational videos and LinkedIn to connect with experts in the field.
Chandrasekhar soon realized that training complex machine-learning models required immense computational power. Without access to supercomputers or high-performance computing clusters needed to train his resource-intensive models, Chandrasekhar explored online initiatives that provided computational resources to projects like his own and simplified his models to increase efficiency.
“A lot of my algorithms took from 30 to 70 hours just to train for a single iteration,” Chandrasekhar said. “It was an excruciating process.”
As Chandrasekhar sifted through clinical notes, he discovered not just patients’ symptoms, but also their life stories. From the social history section, found in many clinical records, he learned about patients’ personal struggles outside of battling cancer.
“It’s touching to see the sacrifices that are made in healthcare and research by patients who are willing to contribute to scientific advancements by providing data about themselves,” Chandrasekhar said.
After fine-tuning his model, Chandrasekhar found gratification in obtaining his first set of statistically valid results. His hard work was recognized by Regeneron Science Talent Search, which was a surreal experience for Chandrasekhar.
“The first thing I felt was disbelief because I didn’t think that it would ever happen,” Chandrasekhar said. “I felt a lot of gratitude towards all the people who had invested their time and efforts in me, like Mr. Iams, who provided advice on my statistical analyses and experimentation.”
Awarded a $2000 prize for his outstanding achievement, Chandrasekhar plans to save a portion for his college tuition. He also plans to invest in future research, like purchasing a graphic processing unit to run more extensive analyses and overcome his current computational limitations.
For Chandrasekhar, being named a Regeneron Scholar is just the beginning. He plans to continue bridging artificial intelligence and healthcare, translating the discoveries made in the lab to the bedside for patients.
“In research, you should pursue your interests boldly,” Chandrasekhar said. “Don’t be afraid to reach out to people, attend conferences and really engage with your field.”
