OncoSum

• 

  We have a ppt too...

OncoSum: Revolutionizing Oncology with AI-Powered Summarization

Inspiration: OncoSum was conceived from the need to streamline the vast amount of information oncologists deal with daily, from patient records and research articles to clinical trial data. Our aim was to create a tool that not only simplifies data management but also enhances decision-making in oncology care.

What It Does: OncoSum fine tunes LLM algorithms to summarize medical conversations, patient histories, and research findings. It presents complex information in concise, actionable insights, enabling oncologists to quickly grasp patient needs and relevant research insights.

How We Built It: Our journey began with assembling a cross-disciplinary team of data scientist and oncology specialist. We leveraged natural language processing (NLP) and machine learning to train our model on a vast corpus of medical literature, ensuring it understands and processes oncology-specific terminologies and contexts accurately.

Challenges We Ran Into: One of the major hurdles was ensuring the AI accurately understood medical jargon and could contextualize patient conversations. Ensuring data privacy and security also presented significant challenges, given the sensitive nature of medical records, so we did not have a lot of training data for this hackathon.

Accomplishments We're Proud Of: We're particularly proud of OncoSum's ability to provide real-time, accurate summaries that have been validated by oncology experts, and the generation of an app.

What We Learned: We learned how to fine tune models and do prompt engineering!

What's Next for OncoSum: Looking forward, we're excited to expand OncoSum's capabilities to include predictive analytics for patient outcomes and treatment efficacy. We aim to integrate OncoSum into electronic health record systems nationwide, making it an indispensable tool for oncologists everywhere. Continuing to refine our model and incorporating more data from the medical community remains a top priority to ensure OncoSum remains at the forefront of oncology care innovation.

All the code, raw data and outputs are on github.

Built With

• huggingface
• llama