Key considerations for the use of AI in healthcare and clinical research

I published an article about key considerations for using AI in healthcare and clinical research. The key points from the article are:

• Medical AI is well-suited to complex input, personalised output and scaling through automation
• However, AI has limitations, and it’s generally better to start with a problem rather than a solution. Its limitations include:
  • AI models are inherently difficult to interpret
  • AI models are only as good as the data they are trained on
  • Some clinical decisions are inappropriate for AI
• Both clinical and machine learning expertise is required for medical AI
• Its not obvious how much data is required (although the more the better)
• There is currently no standardised environment for deployment of AI models
• The regulatory landscape for medical AI is evolving and unclear]

Many of these points are also expanded on in this slideshow presentation (originally delivered as part of this module at Imperial College London).

You can read the full article here.