DinoDoc

• 

  Results of "My back hurts"

• 

  Front page of the application

Inspiration:

In the digital space, finding specific information on the internet can be troublesome; websites offer endless opinions and opposing information. Physical wellbeing is a concern for many citizens who struggle with tech literacy. Their major issue: finding solutions for medical problems without immediate access to a doctor. Meet DinoDoc, a friendly companion, who analyzes your health concerns and determines the best recommendations from your symptoms.

What it does

DinoDoc prompts patients to input any health concerns weighing on their mind. Examples that the patient might input include "I can't put pressure on my knee or bend it straight" or "I bumped my head the other day and now I constantly feel nauseous". After submitting, DinoDoc analyzes through numerous articles, aggregating the data into a concise intuitive summary. Additionally, DinoDoc finds the most likely cause for the patient's condition including the next probable answers and possible treatments. However, if the patient deems the result inaccurate, they can run their concerns with DinoDoc again to display a new report.

How we built it

We utilized Reflex's Python based full-stack framework to build the majority of the website and display the final results. The main concept of Reflex uses a variety of components in combination with State handlers to implement the desired functionality. We combined Gemini and Fetch.AI to power and optimize our information aggregation. Finally, we used Flask to create an API search endpoint that the frontend could access.

Challenges we ran into

One of the major challenges we faced was integrating Reflex's framework with the new Google Gemini API. We faced an import conflict with Reflex and Gemini, which meant we could not import functions inter-module as you would normally. Our solution was to write a Flask API endpoint that served as communication from back to front end. This was an obstacle for us as we were hoping to use Reflex for the whole project. Additionally, there was a steep learning curve to use Reflex's new framework as the documentation lacked all the features we desired. It took many hours of experimentation to figure out bits and pieces of the project.

Accomplishments that we're proud of

As mentioned above, using the Flask API came as an unexpected surprise on the backend. As such, we are very proud of the quick learning we had to do in order to complete the backend. In general, we are very proud of the product we made, as it required quick learning and lots of collaboration.

What we learned

This hackathon introduced us to brand new frameworks/services and we spent a considerable amount of time reading documentation and debugging. However, after spending a considerable amount of time with these libraries, we can now use them in future projects and hacks. Aside from the technical, we learned the value of time management, planning, and how to work under pressure to meet a deadline.

What's next for DinoDoc

We would continue to work on optimizing the integration with Gemini, so the user and API work faster to get the results. Some potential features include: an account system so your history is saved, a language translation feature to improve health literacy, and an integration with medical services. DinoDoc aims to enable communities that lack the time or resources to communicate with doctors to resolve their medical issues. In the future, we hope to collaborate with major pharmaceutical companies to provide the necessary and accurate data to make our application more reliable.

Built With

• fetchapi
• flask
• gemini
• langchain
• python
• reflex