Describe your area of research and/or your area of specialty
My work is in biomedical research analytic approaches designed to work in concert with the expanding landscape of scientific literature, growing at 3.7% annually. The approaches aim to remain agnostic to bibliometric bias, leveraging purely semantic vector representations of LLMs hybridized with network analysis. Our first approach untangles the complexity of scientific jargon, where the same chemical, disease, or exposure may have dozens of synonyms. We demonstrate our hybridization of LLM context embeddings and graph networks in grouping 8,926 potentially hazardous scientific entities into 4,321 unique instances with 88.9% precision. Our second approach builds on the entity grouping to analyze PubMed literature from the last 25 years. Using an in-house LLM with 20% better recall than GPT-4, we extracted and counted potential cancer-causing agents that were either implicated or exculpated in the literature. These counts are then modeled with Bayesian likelihood scores. This approach nominated ~1,600 agents potentially implicated in carcinogenicity not currently in major databases. Our third approach combines LLM and graph network techniques to demonstrate a research center's semantic foresight of its publication output. Using 118,635 historical and contemporary publication abstracts, our algorithm automatically identified 50 self-organized scientific themes through unsupervised learning. Network analysis revealed that the center's semantic foresight—its ability to anticipate emerging research directions—significantly exceeded that of comparable research. Our fourth approach leverages our semantic edge weights in our graph networks to build a classification model trained on expert-labeled publications for impact, novelty, and methodological quality scores, achieving over 90% accuracy in predicting expert consensus while remaining independent of any specific institutional or bibliometric bias.
Describe your involvement within the department and UC San Diego (were/are you involved in student orgs, recreation, certificate programs, internships, etc. outside of curricular requirements?)
Professional Development Student Body JGSC, 2021
Proud former resident MesaHDH Grad Housing
Have you received any outstanding mentorship or guidance during your time in the program that made an impact on your research and/or the trajectory of your career?
My PIs, Professor Parag Katira at SDSU and Professor Ludmil Alexandrov, provided foundational executive insight to help me truly grow into a scientist. Extremely grateful for their leadership and patience. Additional thanks to the kindness and patience from Professor Bernhard Palsson and the kind and professional students in his lab; their guidance and great courses are what taught me the foundational elements of Machine Learning and Deep Learning that led my work's incredible success.
Have you received any awards, graduate grants or fellowships that contributed to your success in the program?
I'm grateful for funding and moral support from both San Diego State University's Advancing Cancer Careers for ExceLlence (ACCEL) and San Diego State University's HealthLINK Center for Transdisciplinary Health Disparities Research (SDSU HealthLINK Center).
What has been your favorite part about your graduate experience in the program?
Being a part of the most cutting edge research, brilliant minds, and fantastic mentorship.
Any thoughts or advice you'd like to share with prospective graduate students?
Finish your coursework, FAST! Get your research project as soon as possible. - That's where you truly learn the most; outside the classroom, thinking and problem solving critically & freely, is how you become a true Scientist.
If you are an Alumni from our program, what is your current role, or what are your career prospects and plans for the future?
I now work as a Research Scientist in AI at the California Medical Innovations Institute, here in San Diego. I also work in the startup space for large scale LLM and AI research analytics.