- Human-centered AI can be defined as putting the Human at the centre of solutions using AI:
- AI adds input.
- Through this synergistic relationship, the impact on humanity can be maximized and optimized and thus justify the partnership and symbiosis of humans and technology.
Regarding clinical decision-making – have we really moved from proof of concept yet, to product in practice ? (2020)
- The “silico” to “in vivo” transition or translation is still pretty early on and recent – we just need more robust evidence to reach the real world.
- Why is that the case ?
- AI has been generating hype for many years now. Why haven’t we made it into the clinical decision-making space yet?
- AI is a complex system: it goes from the design of a model to its deployment to the real world through clinical testing and is then monitored for improvement of the model and error detection.
- Most of our time right now is spent at designing good machine learning practices, focusing on:
- data selection and management.
- model training and tuning.
- model validation.
We have yet to really leave that initial step where we move into clinical deployment and then real-world performance.
- Open questions:
- Why are we stuck at this early stage ?
- Why are we doing all of this ?
- What is the actual problem we’re trying to solve, before we think about ideating and testing and building models that we’ll go ahead and deploy ?
- What is guiding us in terms of practice ?
- Are the objectives of engineers and business colleagues, the same as physicians?
- We need to empathize and understand the people that we’re building these tools for, and really define the clinical need.
Note that the presenter has used the concept of Augmented Intelligence instead of Artificial Intelligence. Please refer to the introductory article of this term.
Ivy Lee, MD. Augmented Artificial Intelligence in Dermatology. Core concepts and clinical decision making. 8th World Congress of Teledermatology, Skin Imaging and AI in Skin diseases – November 2020