The pancreas is one of the hardest organs to see. Tucked behind the stomach in the upper abdomen, surrounded by blood vessels and adjacent structures, it lies in a anatomical neighbourhood so crowded that radiologists — experienced, trained, working from high-resolution MRI scans — can spend minutes locating it before beginning any analysis. During radiation therapy for pancreatic cancer, that difficulty becomes life-or-death: if the radiation beam misses the tumour even slightly, or clips adjacent structures like the duodenum or the spine, the consequences are severe. Rishab Jain was thirteen years old when he decided to solve this problem with deep learning. His AI system, the Pancreatic Cancer Deep Learning System (PCDLS), detected the pancreas with 98.9 percent sensitivity. He was in seventh grade.
Born in Portland, Oregon, in 2005, to parents of Indian origin, Jain grew up in an environment that valued intellectual curiosity and gave him early access to tools for pursuing it. He began learning programming at an age when most children are discovering video games, and discovered a natural affinity for mathematics and logic. When he encountered machine learning — the discipline that enables computers to learn patterns from data rather than following explicit instructions — he understood immediately that it was a tool with medical applications that no one had fully realised yet. Radiotherapy was one of them.
Pancreatic cancer has a five-year survival rate of approximately eleven percent — among the lowest of any common cancer. Late detection is the primary reason: by the time symptoms appear, the cancer has usually spread. But even for patients diagnosed early enough for radiotherapy, the treatment itself carries risks. Radiation beams must be aimed with extreme precision at the tumour while avoiding surrounding tissue. The pancreas, small and irregularly shaped, must be accurately segmented — its boundaries precisely identified — in each cross-sectional MRI slice before treatment can begin. This segmentation, done by hand, is time-consuming, subject to inter-observer variability, and requires constant specialist attention.
Jain's PCDLS (Pancreatic Cancer Deep Learning System) applies convolutional neural networks — a class of deep learning architecture that excels at identifying patterns in visual data — to the problem of automatic pancreas segmentation in MRI scans. The network was trained on a dataset of medical images, learning to recognize the visual features that distinguish pancreatic tissue from surrounding structures. Once trained, it could be applied to new scans in seconds, producing a segmentation map that identified the pancreas with 98.9 percent sensitivity. To put this in context: contemporary state-of-the-art academic systems at the time of Jain's work were achieving sensitivities in the range of 70–80 percent. A thirteen-year-old in Portland, Oregon, had built a system that outperformed them.
In October 2018, Jain presented his work at the Discovery Education 3M Young Scientist Challenge — the nation's premier middle school science competition. He was named America's Top Young Scientist and awarded $25,000. TIME Magazine named him one of its 25 Most Influential Teens of 2018. MIT's Lincoln Laboratory awarded him a minor planet: (12900) Rishabjain now orbits in the asteroid belt, a permanent fixture of the solar system. He received the Gloria Barron Prize for Young Heroes and was awarded the Vincitore Assoluto (Absolute Winner) by the Giuseppe Sciacca Foundation at Pontifical Urban University in Vatican City.
Since his 2018 win, Jain has continued at the intersection of computer science and biology. He gave a TEDxGateway talk in Mumbai in 2020, speaking about curiosity-driven science and the future of AI in medicine. He founded Claisen, a biotechnology startup developing AI-powered treatments for gut health disorders — applying machine learning to a different but equally complex biological system. He has also developed a substantial YouTube presence, running multiple channels with over 160,000 combined subscribers, focused on technology, AI, and science education.
The trajectory of Rishab Jain's career illustrates a pattern that is becoming increasingly recognizable: the young person who encounters machine learning before formal academic training has calcified their thinking about what computers can and cannot do. Without the accumulated prejudices of years of traditional computer science, they approach problems with a kind of naive directness — they ask "can AI do this?" not "should AI do this?" — and they find that the answer is often yes. Pancreatic cancer segmentation. Drug discovery. Space debris tracking. The applications are limitless. What limits them is not the technology but the imagination to see where it can go. Rishab Jain, at thirteen, had that imagination. He still does.
"I trained a convolutional neural network on MRI images and achieved 98.9% sensitivity in pancreas detection. Radiotherapy targeting becomes far safer when the AI can find the organ reliably."— Rishab Jain, on the PCDLS system
The problem: During radiotherapy for pancreatic cancer, radiation beams must be precisely aimed. But the pancreas is small, irregularly shaped, and surrounded by critical structures. Manual segmentation in each MRI slice is slow and variable between radiologists.
Jain's solution — PCDLS: A convolutional neural network (CNN) trained on MRI datasets to automatically identify and segment the pancreas in each cross-sectional image. CNNs learn visual patterns from thousands of examples — edge detection, shape recognition, texture — then apply those learned patterns to new scans.
Result: 98.9% sensitivity — far exceeding contemporary academic benchmarks of 70–80%. The system processes scans in seconds, enabling faster treatment setup and reducing human error. Fewer mis-aimed beams means less collateral radiation damage to healthy tissue.
| Innovator | Country | AI Application | Achievement |
|---|---|---|---|
| Anika Chebrolu | USA (Indian-American) | In-silico molecular docking — COVID-19 spike protein | 3M Young Scientist 2020, $25,000 |
| Amber Yang | USA | Neural network space debris orbit prediction | Intel ISEF $50,000 Young Scientist Award 2017 |
| Kavya Kopparapu | USA (Indian-American) | AI glioblastoma brain tumour grading | Regeneron STS Top 40, GirlsComputingLeague founder |
| Rishab Jain Top Young Scientist | USA (Indian-American) | CNN pancreas segmentation for radiotherapy (PCDLS) | 3M Young Scientist 2018, TIME 25 Most Influential Teens, Minor Planet |
Rishab Jain — TEDxGateway 2020: AI, Curiosity, and the Future of Cancer Treatment
PCDLS — The AI System That Detects Pancreatic Cancer with 98.9% Accuracy
"When I started, the best systems were around 75% accurate. I thought: there must be a better architecture. I tried several, and PCDLS worked."— Rishab Jain, on iterating toward 98.9% sensitivity
"The most powerful thing about machine learning is not what it can do today. It's what it will do when curious people keep asking what comes next."— On the future of AI in medicine
Pancreatic cancer kills approximately 50,000 Americans annually — more than 90% within five years of diagnosis. The primary reason for this catastrophic survival rate is not a lack of treatments. It is late detection and the precision challenges of radiation therapy. Every tool that makes radiotherapy more accurate saves lives. Every improvement in pancreas segmentation — enabling more precise beam targeting — reduces collateral radiation damage and enables higher, more effective doses to be delivered to tumours without destroying surrounding tissue.
Rishab Jain's PCDLS system, built by a thirteen-year-old in Portland, Oregon, demonstrated a principle that the medical AI community had been theorising about: that deep learning could outperform manual segmentation for irregular, hard-to-visualize anatomical structures. When he published his results — 98.9% sensitivity, exceeding state-of-the-art — he was not merely winning a science fair. He was demonstrating, with real data, that AI-assisted radiotherapy planning was not a distant aspiration but a present-day achievability. That demonstration matters. It moves funding, attention, and research effort. A thirteen-year-old pointing at a possible future is, sometimes, the thing that makes that future arrive sooner.