An Overview of Machine Learning: Techniques, Pitfalls and Trust Issues

Bond, R. (Speaker)

Activity: Talk or presentationOral presentation

Description

This talk will present the basics of machine learning including definitions, processes and the typical algorithms used to analyse ECGs. The talk will be accessible to delegates from clinical, technological and industrial backgrounds. Beyond covering the general data science pipeline from data preparation, feature engineering and modelling, the talk will emphasise important issues that data-driven algorithms have and will discuss the importance of integrating domain knowledge. Key issues relating to automation bias (over trusting algorithms), AI explainability and accountability will be discussed. We will also explore the need for ‘explanation user interfaces’ in the next generation of automated ECG interpretation systems. The talk will suggest that providing insight and understanding is arguably more important than providing a mere prediction.
Period28 Apr 2021
Event title45th International Society for Computerized Electrocardiology
Event typeConference
Conference number45th

Keywords

  • machine learning
  • ECG
  • digital health
  • decision making