HippoTree

• 

  Symbol

Inspiration

Flashcards suck. With such a massive library of content and thousands of horrendously long latin names to remember, every new lecture brings hundreds more flashcards to create and remember. By the time you finish that long list of flashcards, the first 80% of flashcards have already vanished from your memory.

How can we allow aspiring medical practitioners to reduce this memory decay? Turning to what we know best, we saw that the shallow learning that flashcards provide is just not enough for a field as vast and complicated as medicine. Memory is retained by having many connections to existing topics so even if the brain blanks out, you know the tracks to that lost point of knowledge. Medicine is about connected reasoning, not isolated facts.

We identified a gap: medical students have little chance to practice real “presentations” — essentially, diagnosing patients. With AI that can both mimic a human patient and recall knowledge without forgetting, our breakthrough was to pair a mock patient with an all-knowing library, giving students realistic, repeatable diagnosis practice.

What it does

HippoTree is a real-time AI patient simulator designed to generate realistic responses and provide university students with practical clinical experience. The platform is capable of creating dynamic flowcharts with color-coded labels to map out all possible outcomes of a given symptom. For example, if a student is studying sore throats, the AI can present associated symptoms such as dry cough or productive cough, while generating precise, structured flowcharts that guide the learner toward identifying the underlying issue.

How we built it

Being First-Years with less than an integer year worth of coding experience between the 5 of us, our plan was to use our vision to mold ChatGPT and Claude AI LLM models into writing code for us. By splitting the project into traditional front-end and back-end sections, our group ran our AI models overtime in crafting the infrastructure to make this project a reality.

Challenges we ran into

Just a few of the challenges we ran into include finding an idea for an app we could code which students studying medicine would actually want and use. We managed to overcome this hurdle by finding a group of post-graduate medical students and interviewing them to find flaws in the way med students study and creating an app to solve those issues. Another issue we ran into was the implementation of an AI patient into our code, through research and many hours of study we managed to achieve this through constant debugging and code streamlining.

We also ran into challenges around the code itself. Since we have literally no experience with any of these languages being used we had to learn on the fly and problem solve with our GPTs to actually create usable code. We needed to be highly specific with our prompts and ensure that features we wanted to keep still existed after GPT rewrote it, essentially fighting in the dark.

Accomplishments that we're proud of

We are proud of the product we have developed within such a short time frame and are confident in its value for enhancing the way medical students study. Our team successfully created a complex program that enables students to engage in practical interactions and gain meaningful information. Completing such a detailed project under significant time constraints is an achievement in itself. Despite the demanding process of planning, coding, recording, and refining, we were able to finalize a workable prototype in less than three days.

What we learned

Throughout the hackathon, we learned the importance of balancing both planning and coding. By dedicating the first day to careful preparation, we were able to construct a comprehensive plan that guided and supported us over the following days. Although the coding process was challenging, we gained valuable experience in creating a user interface, designing a logo, integrating AI, debugging, how to effectively prompt AI, and more. These are invaluable skills that will greatly benefit our future. Thus, the hackathon was not only an enjoyable experience but also a highly educational opportunity.

What's next for HippoTree

In the future, we plan to integrate more sophisticated AI into HippoTree. With this addition, we will be able to generate a limitless range of solutions, offering medical students a wider variety of choices. By incorporating AI, HippoTree will deliver accurate, up-to-date responses that enhance the realism of student interactions. This advancement will help learners engage in more authentic conversations, better preparing them for real-world clinical situations.

Built With

• chatgpt
• css
• html
• java
• tsx
• typescript