Title: Accelerating Rare Genetic Disease Diagnosis with Language Models: A Collaborative Effort at Hack GPT NYC. By Franklin
Mendez
Abstract:
We present a collaborative project undertaken at Hack GPT NYC, aimed at accelerating the diagnosis of rare genetic diseases using advanced language models. With the participation of three dedicated team members, including myself, we developed an innovative tool inspired by the vision of creating a "Google" for genetic diseases. Through the integration of semantic similarity analysis, a ranking system, and prompt engineering, our solution leverages OpenAI's gpt3.5_turbo model, Cohere's rerank product, and a tech stack encompassing HTML, CSS, JavaScript, Python, and Flask. This project showcases the power of collaboration and highlights my specific contribution to the design and implementation of the chatbot prompt engineering module.
Introduction:
Rare genetic diseases present a complex diagnostic challenge, often resulting in significant delays in accurate identification. In our collaborative project at Hack GPT NYC, I actively participated as one of the three dedicated team members. Together, we embarked on the mission to develop a language model-based tool that assists clinicians in the efficient and precise diagnosis of rare genetic diseases. By pooling our expertise and leveraging cutting-edge technologies, our goal was to create a solution that could significantly reduce diagnosis time and positively impact patient well-being.
Solution Approach:
Within our team's collaborative effort, my specific contribution focused on the prompt engineering module of our solution. This module played a crucial role in refining the search for the most probable disease by engaging in dynamic conversations with clinicians. Using HTML, CSS, JavaScript, Python, and Flask, I worked on developing an intuitive and interactive chatbot interface. This chatbot actively inquired about specific symptoms, facilitating the narrowing down of potential diagnoses and ultimately enhancing the accuracy of our tool.
Technological Implementation:
In our project, we seamlessly integrated various technologies to create an efficient and user-friendly diagnostic platform. Leveraging OpenAI's gpt3.5_turbo model for language processing and Cohere's rerank product for disease ranking, we ensured the accuracy and relevance of our tool's outputs. Additionally, my specific contribution involved using HTML, CSS, JavaScript, Python, and Flask to develop the chatbot module, enabling dynamic interactions with clinicians and eliciting specific symptoms for a refined diagnosis process.
Conclusion:
As a collaborative team member in this project at Hack GPT NYC, I am proud of the contribution I made to advancing the diagnosis of rare genetic diseases. Through my specific involvement in the prompt engineering module, we developed an interactive chatbot that enhanced the diagnostic process by refining the search for the most probable disease. Our collaborative effort demonstrates the power of teamwork and highlights the importance of leveraging language models in healthcare innovation. Together, we have created a tool that accelerates the diagnostic process, ultimately improving patient outcomes. I am grateful for the opportunity to contribute to this meaningful project and for the collaborative environment fostered at Hack GPT NYC. Thanks to my team members Md Abedin, and Francisco Requena.
Log in or sign up for Devpost to join the conversation.