As we brainstormed ideas, we were reminded of a close relative waiting in the ER, experiencing unbearable pain. This made us reflect on the inefficiencies in the triage process, where patients are categorized based on their symptoms and injuries. We aim to integrate AI to prioritize patients based on severity and urgency, improving the triage system. This project is especially meaningful to us, as one of our teammates' brother, suffering from a dislocated shoulder, waited for nearly two hours in the ER. The delay in receiving prompt care highlighted the need for a better system to quickly identify and prioritize critical cases.
What it does
SwiftCare is a web application designed to enhance emergency care services by improving the triage process. By integrating pre-trained AI models, we aim to classify and prioritize patients based on the severity and urgency of their symptoms. This allows healthcare providers to respond more efficiently and ensure that critical patients receive immediate attention. Our solution leverages advanced AI to optimize the triage system and ultimately improve patient outcomes.
How we built it
We built SwiftCare using a combination of React, Node.js, HTML, CSS, and integrated Hugging Face’s pre-trained AI models for classifying patients based on their symptoms. The frontend was designed with HTML & CSS to provide an intuitive and responsive user interface, while the backend powered mainly by python handles the AI predictions. React was used to connect the front-end and the back-end.
Challenges we ran into
Initially, we thought Gemini API is completely free and planned to use vertex AI to train our own model. Later, after spending hours, trying to integrate gemini AI studio , we realized that Vertex AI is higher tier and it's a paid version. Even though, this was disheartening , we were able to find ways to make our project into reality by using hugging faces trained AI model. Lastly, it's the obstacles we faced during the hackathon made us motivated and more eager to take that next step to increase the chances of winning awards.
Accomplishments that we're proud of
We are proud of our resilience and determination. Despite the setback of not being able to use Gemini AI Studio, we didn’t let it deter us. We successfully integrated Hugging Face's pre-trained AI models, which allowed us to accomplish what we set out to do. Through this project, we not only learned about AI but also gained hands-on experience in integrating it into real-world applications. This accomplishment gives us the confidence to explore more projects involving AI in the future.
What we learned
We used in react & hugging faces pre trained model AI . As react is based upon javascript , we learned a lot of the functionalities and differences from the programming languages we know. Moreover another interesting thing, we learned was integrating pre trained ai models from hugging faces which is super cool and we can't wait to use it in another projects and leverage it.
What's next for SwiftCare
As the hackathon comes to a close, the project that kept us awake, typing away on our laptops, is far from over. SwiftCare is just beginning. We plan to continue developing it into an app and integrate features like voice commands and image recognition to improve accessibility and usability. While we understand that images may not always be the best solution due to potential risks like disguise or misinterpretation, we are committed to exploring innovative ways to enhance the user experience.
In the future, we aim to expand SwiftCare by offering more comprehensive support for emergency care, integrating AI-driven diagnostics, and providing real-time patient tracking. The journey ahead is long, but we are excited to take SwiftCare to the next level and make it an essential tool in emergency healthcare.
Log in or sign up for Devpost to join the conversation.