The problem Pulse Pal solves

Our app revolutionizes healthcare by streamlining patient record management and pre-visit screening, significantly boosting efficiency and diagnostic precision. Here's how it benefits both doctors and patients:

For Healthcare Providers:

• Time-Saving: Automates compiling fragmented medical records, saving an average of 16 minutes per patient. This allows doctors to focus more on patient care rather than administrative tasks. It also summarizes radiology reports!
• Pre-visit Screening: Utilizes remote photoplethysmography (rPPG) technology for early detection of conditions like cardiovascular diseases by analyzing heart rate variability metrics. This screening ensures doctors can prioritize patients based on urgency and have critical health insights before the consultation begins.
• AI-Powered Chatbot: Designed for doctors, this feature collects patient symptoms and inquiries beforehand, enabling informed decision-making and more personalized care during consultations.

For Patients:

• Noninvasive Screening: The app's use of rPPG technology offers a convenient, noninvasive way for patients to perform preliminary health checks, potentially identifying serious health issues early.
• Dynamic Health Monitoring: Enables patients to monitor changes in their health over time, alerting both them and their healthcare provider to any signs of deterioration for timely intervention.

Challenges we ran into

Integration Challenges

• Difficulties integrating diverse components: UI, backend services, generation models, and rPPG technology module.
• Communication issues among different parts of the app.
• Solved through modifying security privileges.

External Dependencies

• Reliance on external services like Hugging Face for AI chatbot features.
• Encountered unexpected crashes due to server overload and unexpected maintenance.
• Used library work around and Hugging face worked in the end.

Verification Delays and Lack of Access to Research Materials

• 3 day verification process for model data, which is impractical in a hackathon setting, and scarcity of existing libraries
• There was difficulty in accessing research data which comprised the majority of our data needs.
• Solved by scraping data and converting all files to text to run model.