SepsisX

• 

  SepsisX —it’s a lifesaving innovation

Inspiration: Sepsis is a life-threatening condition that requires early detection and immediate intervention. Many hospitals struggle with late diagnoses, which significantly impact patient survival rates. Our team was motivated to develop SepsisX to leverage AI for early detection, enabling faster clinical decision-making and saving lives.

What It Does: SepsisX is an AI-powered prediction system that integrates with FHIR-based electronic health records (EHR) to assess sepsis risk in real time. It uses machine learning models to analyze patient vitals, lab results, and medical history, providing clinicians with risk scores, early warning alerts, and explainable AI insights.

How We Built It: We developed SepsisX with a FastAPI backend for high-performance processing and a Next.js frontend for an intuitive user experience. The system integrates with FHIR APIs for real-time patient data retrieval, and machine learning models trained on medical datasets enhance prediction accuracy. Redis and Celery enable real-time event processing, while Docker and Kubernetes ensure seamless deployment on cloud platforms like AWS and GCP.

Challenges We Ran Into: One of the biggest challenges was ensuring real-time data streaming from EHR systems while maintaining accuracy and speed. Optimizing AI models to reduce false positives required extensive fine-tuning. Implementing secure authentication for HIPAA compliance and designing an intuitive user interface to make AI insights actionable also posed significant technical hurdles.

Accomplishments That We’re Proud Of: We successfully built an AI-powered system that can predict sepsis risk with high accuracy and provide real-time alerts. The seamless integration of FHIR data with our machine learning models was a major achievement. Additionally, the development of an intuitive clinician dashboard with real-time patient insights was a significant milestone.

What We Learned: Throughout this project, we gained deeper insights into FHIR interoperability, healthcare data security, and optimizing machine learning models for real-time predictions. We also learned how to enhance AI explainability using SHAP and LIME to ensure trust and usability for healthcare professionals.

What’s Next for SepsisX: Moving forward, we plan to integrate wearable IoT devices for continuous patient monitoring, expand our model with federated learning for improved accuracy across hospitals, and develop a mobile application for real-time alerts. We aim to make SepsisX a comprehensive, AI-driven tool that transforms how sepsis is detected and managed in hospitals.

Built With

• aws/gcp
• backend:-fastapi-(python)
• celery
• chakra-ui
• ci/cd
• github
• hapi-fhir-?-deployment:-docker
• kubernetes
• lime-?-fhir-integration:-google-cloud-healthcare-api
• nginx
• oauth2-?-frontend:-next.js-(react)
• postgresql
• pytorch
• recharts-?-machine-learning:-tensorflow
• redis
• shap
• xgboost