About
The overall topic is medication management, and the problem is that for people who have to self-inject estradiol or testosterone multiple times a month, there are a lot of things to manually keep track of for optimal practice -- such as hormone levels from past lab results, tracking past injection sites for easy site rotation, adaptive reminders for next injection, and possibly others, such as mood tracking across cycles.
In addition, our solution is aimed at giving autonomy to people in an increasingly politically unstable environment, with many actively and passively hostile clinicians who may not have their patients’ best interest at heart. Further, the vast majority of doctors lack the experience and knowledge to properly care for trans patients, and often prescribe doses that go against well-established guidelines, such as those from UCSF. This app would ensure people have a baseline level of HRT advice, in leiu of a qualified doctor.
The goal is creating an app for managing and unifying a lot of the steps in self medication in one place. This app will be minimalist and as simple as it can be, and will be customizable to the users’ needs, avoiding unnecessary bloat like LLM chatbots.
While there are recent commercial attempts like ChatGPT Health that can feed LLMs your EHR and other health data, LLMs are notoriously bad at healthcare [Yang et al, Tian et al] and can lead to life-threatening outcomes. This app would be much safer alternative by using models calibrated with patients’ data.
----
hi, I’m Ashley Shin, a CSE PhD student at UC San Diego. My main research interests are in applied machine learning — retrieval and recommender systems — and in the past, I’ve worked on biomedical NLP at the NIH. This is a design project for improving medication management and levels monitoring for hormone therpy, eg estradiol.
References: Yifan Yang, Qiao Jin, … Zhiyong Lu. 2025. Beyond Multiple-Choice Accuracy: Real-World Challenges of Implementing Large Language Models in Healthcare. Annual Review Biomedical Data Science. 8:305-316. https://doi.org/10.1146/annurev-biodatasci-103123-094851 Shubo Tian, Qiao Jin, … Zhiyong Lu, Opportunities and challenges for ChatGPT and large language models in biomedicine and health, Briefings in Bioinformatics, Volume 25, Issue 1, January 2024, bbad493, https://doi.org/10.1093/bib/bbad493