Age: 17Cupertino, CA
Project Title: Decoding Neural Networks: Novel Computational Methods to Discover Anti-Tumor B Cell Receptor Binding Motifs
Cynthia’s study is the first to provide a framework for interpreting the motif patterns learned by deep learning models trained on protein sequence data. Deep neural networks have achieved great success in diagnosing diseases, but they remain black boxes: scientists are often unable to clearly explain how a model arrives at its decision or which features matter most. To address this, Cynthia developed computational methods to uncover the patterns learned by a deep neural network that predicts cancer types based on B cell receptor (BCR) sequences. By decoding this model, she identified and validated 65 tumor-specific BCR binding motifs for 13 cancer types, a discovery that could guide future synthesis of antibody drugs for targeted cancer treatments.
Cynthia Chen is a rising high school senior from Cupertino, California. She is deeply interested in harnessing the power of math and computer science to solve real-world problems. In particular, Cynthia is fascinated by computational biology research and has devoted a lot of time to it over the past three years. She is truly honored and humbled to be named a 2019 Davidson Fellow Laureate. Receiving this award means a lot to her, and she is extremely grateful to the Davidson Institute for their support in her academic and scientific pursuits.
Cynthia’s research addresses the current black-box nature of deep learning. Although artificial intelligence and deep learning models are reinventing the field of computational biology, scientists have difficulty understanding the mechanisms behind a model’s decision or explaining the behavior of an AI system. To tackle this problem, she developed computational methods to decode and interpret the sequence patterns learned by a deep neural network that predicts cancer types based on B cell receptor (BCR) sequences. By decoding this model, Cynthia discovered 65 anti-tumor BCR motif signatures of 13 cancer types, information that could help synthesize targeted antibody drugs for cancer treatments. Fundamentally, she believes that creating transparent and interpretable AI is crucial to increasing the safety, reliability, and acceptance of such technologies and applying them to social good.
Cynthia is extremely grateful for her mentors, teachers, and family—her research would not be possible without their continuous support and guidance. She would like to thank her mentors, Dr. Sherlock Hu and Professor Shirley Liu of Harvard University, for guiding her through the entire journey, taking time out of their busy schedules to meet with her, and providing valuable advice. Throughout her project, Cynthia encountered many challenges, from identifying effective methods for data analysis to generalizing her pipeline for different types of cancers, and most notably dealing with the weak correlations in validation analyses. With the encouragement and guidance from her mentors, Cynthia overcame these roadblocks by devising creative solutions such as novel ranking, clustering, and visualization algorithms through numerous trials and errors. She is thankful for all the Liu Lab members—being able to work with such a group of amazing scientists is truly a humbling experience. Cynthia would also like to thank Mr. Chris Spenner, her research teacher, who taught her what it means to conduct rigorous research and offered unique insights and perspectives on how to present her work effectively. She is indebted to every teacher who has provided her not only knowledge, but also inspiration and encouragement. Lastly, she is forever grateful for the unconditional love and support from her family.
Cynthia’s research is an important first step towards a better understanding of deep learning models that are used in cancer research. The computational pipeline she developed not only provides a novel framework for decoding sequence-based deep neural networks but also identifies the genetic signatures of multiple types of cancers. Cynthia’s work could help reduce the time and cost of developing more effective and targeted cancer treatments for patients.
While conducting research, Cynthia drew heavily upon the knowledge and skills that she learned from her school classes and online courses. By challenging herself with a rigorous curriculum including advanced courses in math and computer science, such as Linear Algebra, Multivariate Calculus and Differential Equations, Discrete Math, and Expert Systems, she laid a strong foundation for her research. In addition to her school curriculum, Cynthia self-studied several online courses in machine learning, computer vision, and biostatistics, which have greatly helped her understand the fundamental concepts within artificial intelligence and computational biology. She is very thankful for these educational opportunities that have expanded her academic horizons and opened up possibilities of research.
Cynthia is very passionate about increasing accessibility to research opportunities for underprivileged students. In 2016, she started Opportunity X (OpportunityX.org), a student-led nonprofit that runs weekly research programs at underrepresented middle schools. With the hard work of the team, Opportunity X has held 180+ workshops at 9 middle schools across multiple states. Cynthia has also been recognized as a Research Science Institute Scholar, Siemens Semifinalist, Third Award winner at Intel ISEF, and Broadcom MASTERS Finalist. In addition to research, she enjoys solving challenging math problems and traveling with her school team to various math competitions across the nation. At school, she serves as the president of the Research Club and AI Club as well as a content editor for the student-run science journal. Besides STEM, Cynthia loves making art and drawing the world around her, which gives her a creative space to think and imagine beyond boundaries. For the last 7 years, she has also played volleyball in school and club teams. In her spare time, she likes to spend time with her family and dabble in culinary missions, including making smoothies and baking yummy desserts. In the future, Cynthia hopes to further explore the field of computational biology and combine her love for math, computer science, and biology to develop innovative technologies for combating terminal diseases.
Where do you see yourself in 10 years?
Working and collaborating at the intersection of multiple scientific fields to search for creative solutions that improve cancer treatment.
If you could have dinner with the five most interesting people in the world, living or dead, who would they be?
Fei-Fei Li, Leonardo da Vinci, Marie Curie, Elon Musk, Alan Turing
If you could be on any TV show, which one would it be?
South Bay Youth Gain Prestigious Davidson Institute Scholarships
CUPERTINO TEEN AWARDED $50,000 FOR DEVELOPING METHOD TO PREDICT CANCER TYPES FOR USE IN FUTURE TARGETED TREATMENTS
Cynthia Chen to be Named a 2019 Davidson Fellow Scholarship Winner
Cupertino, Calif. – The Davidson Institute for Talent Development has announced the 2019 Davidson Fellows Scholarship winners. Among the honorees is 17-year-old Cynthia Chen of Cupertino. Chen won a $50,000 scholarship for her project, Decoding Neural Networks: Novel Computational Methods to Discover Anti-Tumor B Cell Receptor Binding Motifs. She is one of only 20 students across the country to be recognized as a scholarship winner.
“I am truly honored and humbled to be named a 2019 Davidson Fellow,” said Chen. “Receiving this award means a lot to me, and I am extremely grateful to the Davidson Institute for their support of my academic and scientific pursuits.”
Chen’s research is an important first step towards a better understanding of deep learning models that are used in cancer research. The computational pipeline she developed not only provides a novel framework for decoding sequence-based deep neural networks but also identifies the genetic signatures of multiple types of cancers. Chen’s work could help reduce the time and cost of developing more effective and targeted cancer treatments for patients.
Chen is the founder of Opportunity X (OpportunityX.org), a student-led nonprofit that runs weekly research programs at underrepresented middle schools and has held 180+ workshops at nine middle schools across multiple states. Outside of science, Chen loves art and drawing the world around her.
Click here to download the full press release
Click the image to download hi-res photos of Cynthia:
The following disclosure is provided pursuant to Nevada Revised Statutes (NRS) 598.1305:The Davidson Institute for Talent Development is a Nevada non-profit corporation which is recognized by the Internal Revenue Service as a 501(c)3 tax-exempt private operating foundation. We are dedicated to supporting the intellectual and social development of profoundly gifted students age 18 and under through a variety of programs. Contributions are tax deductible.
Profoundly gifted students are those who score in the 99.9th percentile on IQ and achievement tests. Read more about this population in this article.