Haoli is a student from the United States who attended the Summer STEM Institute (SSI). Haoli is interested in the intersection of medicine, technology, and biology. To explore and share his passion for these fields, Haoli has organized and led biology tutoring classes, qualified as a USA Biology Olympiad (USABO) Semifinalist, and developed an edtech platform for intuitive course delivery. During SSI, Haoli worked on a project, “Prostate Lesion Detection and Salient Feature Assessment Using Zone-Based Classifiers,” that uses deep learning to diagnose prostate cancer lesions. You can read his full paper here.
Ever since I was little, I’ve always been interested in how life works, from how cancer forms to how the stomach digests food. As I grew older, I began investigating these topics. I became interested in interdisciplinary sciences and have found the cross-over between biology and computer science especially interesting.
In my freshman year of high school, I started computer science and took my first computer science class on Java and object-oriented programming. From there, I was able to take at least one computer science pretty much every year in high school, from Java to Python. I also competed in the USA Computing Olympiad (USACO), where my interest in computer science sparked.
For the past few years, I have learned about CRISPR systems through Science Olympiad. Recently, I read an article about a new technology called retrons, which is very similar to CRISPR and is part of the bacterial immune system. They can be used for gene-editing, which I think has a lot of potential in the future. I also recently became interested in the research being done at the Stanford Prakash lab. I love their idea of “frugal science” where you take advanced medical topics and turn them into accessible health resources, like toys, to people all over the world. I think this is a really important area we should concentrate on, especially during the pandemic.
For the past four years, I’ve been involved in the Science Olympiad where I was able to pursue my interests in biology, computer science, and engineering. For the past two years, I have been the captain of my team and have led and motivated my team to succeed at numerous national tournaments.
Outside of Science Olympiad, I tutor students. I found that quarantine really restricted the amount of learning that could happen within my school and community. During quarantine, I was able to learn front end development languages, and I competed in various global hackathons with a few friends. We developed Modulus, an edtech startup that allows for intuitive course delivery to continue education during this time. Modulus is a peer-to-peer course delivery platform where students and teachers can learn, teach, and deliver content. So far, we’ve partnered with around 25 education nonprofits and have hosted over 2,000 users on five continents. That’s something I’m really excited about, and you can learn more about the platform here.
For the past seven years, I’ve played the viola. I really enjoy making music and listening to music, all the way from classical genres to lo-fi. I also have worked on engineering projects, such as building model planes. In Science Olympiad, I participate in an event called Wright Stuff, where you essentially build a model plane. I really enjoy tinkering, designing, and just experimenting. I’ve learned a lot from the experimental process.
I've always wanted to, but I've never found the opportunity. SSI opened up this whole new world of research for me. I had such a great first experience and I look forward to my next research opportunity.
Over the summer, I conducted medical imaging research and used machine learning to detect prostate cancer from MRI scans. MRIs have played an increasingly important role in the detection of prostate cancer lesions, but radiologists have a wide range of experiences and training, which can lead to inaccurate results and human error. I implemented 5 different machine learning classifiers to study three zones of the prostate. I trained a deep learning classifier to help validate a radiologist’s predictions, increased the interpretability of these models through saliency models, and paved the way for future investigation to better understand hidden features and biomarkers for prostate cancer detection using MRIs.
As I mentioned before, I’m very interested in biology and computer science, so I wanted to do a project that involved both. That’s why I was interested in medical imaging; it incorporates biology, data science, and cancer research. My mentor introduced me to computer vision and I just jumped straight in. I found a prostate cancer dataset, did more background research into gaps in the field, and came up with my project from there.
That’s a funny story. Two weeks into my project, I found out that I had data leakage. I was testing my machine learning classifiers on the same set of data that I used to train it on, which was fundamentally wrong. I had to brainstorm potential solutions with my mentor, and we came up with a data augmentation method to generate more negative samples so that different images could be used between the training and test sets, thus preventing leakage.
I think it was embracing uncertainty as a problem solver. The most exciting part for me was finding out the results. Seeing all your work compiled in the end, especially the performance graphs, is very rewarding. I saw that over time, I had higher AUCs, meaning my classifiers got better at the task I was trying to complete. I had an AUC of 0.73 in the beginning, but after augmenting the data and training the model, it came up to 0.98, which was absolutely amazing.
Learning programming and data science was really valuable for me. Even though I took several computer science courses before, I usually learned about algorithmic development, like how to reduce Big O and how to make sure everything is as efficient as possible. From SSI, I was able to learn data science and how to utilize different Python libraries, which is something that you don't really learn in school. Learning about all these libraries and how to apply them to my project was very valuable.
I also learned how to be resilient in the face of challenges, especially when I stayed up until 2 AM trying to finish different aspects of my project and kept running into more and more bugs. I learned that Stack Overflow is your best friend, and I learned how to debug. Also, I learned how to write a scientific paper. Writing the abstract was especially challenging because I had to condense my entire paper into one little paragraph, so I worked with an SSI mentor to plan out my abstract.
It was definitely meeting lots of people throughout the nation. That's something that I don't really get to do very often since my only other opportunity of meeting people is through Science Olympiad national tournaments. I really enjoyed being able to relate to other students while bonding over a similar passion for scientific research and development. This is a memory I’ll carry throughout my life, and I’m still in contact with some of these students. Obviously, you learn so much from SSI, but people are kind of what make SSI.
I’m really looking forward to the spring. I have an internship at a heart clinic, where I will shadow doctors and learn how to apply EKGs and all sorts of other things. I also have an interview with a Stanford bioengineering startup coming up. Over the summer, I’m hoping to conduct research again and further the research that I’ve already been doing.
During SSI, learning about data science and its applications in the medical field was really important for me, and it really inspired me to be open-minded to different tasks and to remain curious. In college, I hope to major in biomedical engineering and explore everything this diverse field has to offer, from tissue engineering to biocomputation to medical imaging. As of now, I envision two potential paths after that. Hopefully, I’ll either go into research and development at a medical device company or practice medicine as a radiologist or cardiologist.