Inspiration
During space travel, personal health and safety matters more than ever before. Having a personalized tool that tracks different health metrics like sleep, nutrition, and exercise can go a long way in making the journey more enjoyable. Our team wanted to address this specific challenge with EVE.
What it does
By taking user inputs about body specifications like weight, age, height and gender, this tool is able to perform calculations to see the individual’s required nutrients. With this information, a large language model will then take this to create a personalized meal plan for the individual to follow. The individual will also use this program to keep track of their activity, and sleep, comparing it to their recommended metrics. Finally, the user can upload this data onto their watch, which will provide a minimalist and accessible way to access important information.
How we built it
We followed a modular approach when building this project. We started by designing a basic dashboard with different pages using React.js. Then, we focused on the landing page, ensuring the project properly collected and stored user metrics like name, age, sex, height, and weight. After ensuring the landing page works consistently, we moved on to the meat and potatoes of the project; the dashboard. We ensured to have similarities among the pages for more efficient programming and aesthetics. We started by designing the elements on the nutrition page and modifying them for the other pages as well. At the same time, another one of our team members worked on building the hardware side of the project, making the Arduino. We linked the Arduino to our project and moved on to integrating the LLM into this project. We used Gemini's free API keys to personalize the recommendations for each user and once everything was functional, we moved on to making the entire project aesthetically pleasing.
Challenges we ran into
The first challenge was finding a way to communicate between the web page and the Arduino in order to take those inputs and display them on the LCD screen. This was resolved by writing a program that would take all of the information collected and store it into a single string packet to be sent out the the arduino over Serial. The second challenge was setting up the Large Language Model. Determining a way to format the prompts to the LLM in order to give us the exact prompts that we needed was more difficult than expected. Eventually, after testing, we found a prompt that was simple enough to run while also encompassing all of our parameters.
Accomplishments that we're proud of
Setting up the Arduino and connecting it to our project was an exciting experience and taught us the importance of perseverance (specifically the LCD display) Getting the LLM to work and personalize the meal plan was another fun part of this project.
What we learned
We learned about integrating LLMs into custom programs, and using the power of these models to create even deeper customization than the conversation models. We also learned a lot about website design, and making an intuitive user experience. Finally, we learned about the communication between different programs in the computer, and how we can send and receive information across them.
What's next for EVE
The next step for EVE is scalability to make the project even more user friendly. Currently, the project doesn't account for user allergies or dietary restrictions for the meal plan. It also does not personalize the exercise plan based on user goals and plans. Having the LLM model account other metrics about the user into the project would make it more inclusive and personalized for everyone's growth.
Log in or sign up for Devpost to join the conversation.