Atrial Fibrillation Classification

Accurate classification and actionable management of AFib

UC Berkeley MIDS Capstone Project

Welcome to our MIDS Capstone website.

Project Overview

Atrial fibrillation, or AFib, is an irregular heartbeat that can cause painful episodes, significantly impacting daily life, and leading to severe complications such as stroke, heart failure, and reduced overall well-being. This makes accurate classification essential for timely, life-saving interventions.

Our team has developed an innovative solution that addresses AFib classification by leveraging LSTM and K-Means smoothing to classify heartbeat rhythms as AFib or normal using just ECG data, allowing us to generate personalized insights for a wide audience. Users can use our product, Atrial Insights, to create a personal dashboard designed to help them better understand their atrial fibrillation episodes using real data from their monitoring devices.

Quick Links

• Problem Statement
• Our Solution
• Explainability
• Technical Details
• Team Members
• Documentation

Project Highlights

Accurate classification

F1 score of 94%, balanced across classes

More information, faster

Able to leverage commonly available smartwatch data for insights

Empowering users

Demystifies the diagnostic process