Stanford Seminar - Deep Learning for Medical Diagnoses

This seminar examines deep learning for clinical diagnosis through examples in arrhythmia detection, chest radiograph interpretation, and knee MRI. It discusses dataset curation, model architectures, evaluation, interpretability, external validation, and translation into clinical decision support.

Syllabus

Introduction.
Traditional Model of Clinical.
Diagnostic Procedure with Al.
Information Gathering Step.
Diagnostic Testing with Al.
Machine Learning Framework.
Next Paradigm shift?.
Arrhythmia detection.
Future of continuous monitoring.
Holter Monitor.
Amount of data capture.
Automated Detection Challenges.
Previous Approaches.
Setup.
Network Architecture.
Residual Networks.
Wide ResNets.
Dataset - Test Set.
Medical Errors.
Chest Radiograph Interpretation.
Chest X-ray exam.
Detecting Abnormalities.
X-ray findings of pneumonia.
Detecting Pneumonia.
DenseNets.
Dataset - Train Set.
Evaluation -- Limitations.
Class Activation Maps.
Interpretations.
Future of diagnostic access.
Diagnosis Clinical Decision Support.
Knee MR.
Input.
Dataset Training on.
Interpretability.
External Validation.
Diagnostic Future with Al.