Inspiration
I was inspired to create CareNotesAI after observing the long wait times in Canadian clinics, which can often exceed 2 hours. I spoke with doctor friends and my family doctor, and they shared that one of the major contributors to these delays is the time-consuming process of documenting patient notes. This administrative burden not only creates backlogs but also limits the time doctors can spend with their patients. I wanted to create a solution to help alleviate this issue by streamlining the documentation process.
What it does
CareNotesAI transforms concise doctor summaries into detailed, structured patient notes. The tool uses AI to generate comprehensive outputs, including diagnosis, treatment plans, medication instructions, and follow-up recommendations. It is designed to save time, enhance accuracy, and reduce the administrative workload for healthcare professionals, allowing them to focus more on patient care.
How I built it
I built CareNotesAI using Streamlit to create an easy-to-use web interface and integrated Google AI Studio's Generative APIs for the core AI functionality. I refined the prompts to guide the AI model in producing structured, actionable patient notes. To ensure clinical relevance and usability, I added post-processing logic that guarantees all necessary sections are included and formatted according to healthcare standards. Data security was a priority, and I implemented best practices to comply with privacy regulations like PIPEDA.
Challenges I ran into
Data Privacy: Adhering to healthcare privacy standards made me decide against adding an audio recording feature to the app, as it could raise concerns about sensitive patient data security. Ensuring that the app complies with privacy regulations like PIPEDA was a key priority.
Customization: Balancing a generalized solution with the flexibility to adapt to specific healthcare workflows and requirements was challenging. I fine-tuned the AI prompts to align with the SOAP (Subjective, Objective, Assessment, Plan) documentation format, ensuring the generated notes met clinical standards while being adaptable to diverse needs.
Accomplishments that I'm proud of
- Developing a tool that addresses a real-world problem and reduces the administrative burden for doctors.
- Creating a user-friendly interface that is easy for healthcare professionals to adopt.
- Building a solution that has the potential to improve clinical workflows and reduce patient wait times significantly.
What I learned
Through this project, I learned how important it is to design solutions that align with the end-user's needs and workflows. I gained a deeper understanding of the challenges faced by healthcare professionals and the critical role documentation plays in patient care. On a technical level, I learned how to fine-tune generative AI models to produce outputs that are relevant, actionable, and compliant with real-world standards.
What's next for CareNotesAI
- EHR Integration: I plan to integrate CareNotesAI with Electronic Health Record (EHR) systems to automate data entry and export.
- Template Library: I aim to expand the app to include customizable templates for various documentation needs, such as consent forms and discharge summaries.
- Provincial Adaptations: I want to tailor the outputs to meet the specific guidelines and requirements of different provinces in Canada.
- Deployment: My goal is to scale the application for use in clinics across Canada and eventually internationally.
- Feedback Loop: I will work with healthcare professionals to gather feedback and refine the tool for better usability and performance.
Log in or sign up for Devpost to join the conversation.