Analyze a research paper on an AI-based diagnostic system for COVID-19 using cough recordings. Explore the key assumptions, system architecture, and biomarker models used in this innovative approach. Delve into the ResNet50 architecture, MIT Open Voice Brain Model, and various biomarkers for muscular degradation, vocal cords, sentiment, and respiratory tract. Compare baseline and pretrained models, discuss future work, and gain insights into the potential of AI-powered audio diagnostics in the fight against the pandemic. Join the conversation on The Sound of AI Slack community to further discuss this cutting-edge research and its implications for quick and reliable COVID-19 diagnosis.
Covid-19 and Cough - This AI Predicts if You Have the Infection - Paper Explained
Valerio Velardo - The Sound of AI via YouTube
Overview
Syllabus
Intro
Covid-19 diagnostic problems
Key assumption
Covid-19 diagnosis system
Benefits of Al audio diagnosis
How does the diagnosis system work?
How good is the system?
Use cases
Creating a Covid-19 Cough Dataset
Filtered Covid-19 Cough Dataset
Preprocessing steps
The architecture
ResNet50
What are the biomarkers models?
An architectural problem
MIT Open Voice Brain Model
Intuition
Injecting domain knowledge
Muscular degradation biomarker
Muscular degradation: The math
Why introducing muscular degradation?
Vocal cords biomarker
Sentiment biomarker
Lungs and respiratory tract biomarker
Baseline vs pretrained model
Future work
What did we learn?
Join the discussion!
Taught by
Valerio Velardo - The Sound of AI
