Pinpoint Ai

50 patients per emergency department per year die of misdiagnoses. Pinpoint Ai references specific medical knowledge of top ER misdiagnoses, pointing the medical field in a safer direction.

Comment

Inspiration//

The inspiration behind PinpointAI comes from the number of misdiagnoses in the emergency room every year. Misdiagnoses in the face of an emergency can be detrimental as 50 people per emergency room per year die due to this. At PinpointAI, we want to utilize AI to help medical workers "pinpoint" to a more accurate diagnosis. Existing AI like ChatGPT has its flaws it brings in all the information from the web, including inaccurate ones; however, PinpointAI looks at medically reviewed journals and past research to diagnose symptoms.

What it does

Pinpoint AI takes in accurate and medically endorsed research and reviews journals to diagnose a patient based on their symptoms.

How we built it

We used langchain pdf vector store, openai api, and streamlit. To create the facet of our app that allows medical data input, we used langchain to take an upload and vectorize it for openai to reference. We used streamlit as a frontend for our inputted emergency room queries and to render the langchain upload.

Challenges we ran into

Time was a big constraint for PinpointAI.

Accomplishments that we're proud of

We were able to successfully display an accurate diagnosis on PinpointAI whereas ChatGPT was not able to come up with an accurate response to the same question.

What we learned

We learned how to utilize langchain pdf loader along with openai api, deploy an app on streamlit, develop our idea, and drive it into technical implementation.

What's next for Pinpoint Ai

Right now, only a chat feature is available for Pinpoint AI. We understand that time is expensive in emergency rooms, therefore, the future of Pinpoint AI consists of including live transcription options, and even an image option where doctors can take pictures of symptoms and get back accurate diagnoses. If we can get PinpointAI in hospitals and emergency rooms, security is something that we need to focus on as we have to abide by HIPAA laws to keep patient data confidential. Having confidential access to hospital data would also help further the accuracy of Pinpoint AI as it would be able to use patient symptoms that have previously not been recorded in medical journals to diagnose another patient with similar symptoms.

Built With

Share this project:

Updates