Editor’s Choice Pick: “A Novel Approach for Early Detection of Alzheimer’s Disease” Read about one of our Editor’s Choice picks and the student author at the Journal of Emerging Investigators.

Meet the Student Author

Aryan is an 8th grader at Solorsano Middle School in California. Since elementary school he has been interested in science fairs and projects, building his first potato battery in 4th grade. Aryan has previously published with JEI where he used Drosophila to look at the ability of various natural compounds to reduce tumor growth.

These days, Aryan is primarily interested in coding using his skills in Python to develop various machine learning models and transfer learning approaches to address important issues. He hopes to be able to continue research throughout high school and beyond. Aryan also says that publishing with JEI has been a great experience as he learned how to present his work in a structured manner that made it accessible to a more general audience.

Learn More About Aryan’s Research

Aryan was interested in developing a way to detect Alzheimer’s earlier in patients via MRIs. He became interested in this field of research after a close friend’s grandfather was diagnosed with Alzheimer’s and he saw how difficult it was not only for the patient, but the rest of the family.

After reading articles about Alzheimer’s, both general overviews about the disease and its detection, Aryan decided to create a custom convoluted neural network deep learning model that would extract features from MRIs of patients with Alzheimer’s disease at various stages. He was particularly interested in being able to identify features that are present in the early stages of Alzheimer’s as that provides the best chance for mitigating and/or delaying the onset of some symptoms.

Aryan trained his model across two different image sets, then compared its performance against two other previously developed models. He had predicted that his model would perform better because it contained more layers, but what he actually saw was that while his model performed well, it may have had too many layers. The extra layers could lead to overfitting of the data since the MRIs he validated his model on may have not had enough features for his model to extract.

Please check out Aryan’s full manuscript to see why it was selected as an Editor’s Choice manuscript.

The material on this page was prepared by Kari Mattison, JEI Editor in Chief. Aryan Ganesh provided the photo and personal biography which was edited lightly for clarity.