Thoughts on Artificial Intelligence in Health (last updated 27 Feb 2016)
Key points:
- Current edge in AI is self-teaching systems
- Data scientists offer a unique edge to information-dense, complex systems - make better decisions in the oncoming period of disruption.
- The use of AI in Healthcare will aim to reduce the downside of the critical human element (eg fatigue, mistakes, gaps in knowledge) and augment the benefits (more patient-doctor time, comprehension / observation of complex emotional situations, comfort, human touch and compassion)
- 26 Feb 2016: Google AI group, DeepMind, launches DeepMind Health with two initial apps: Streams, to find patients at high risk of acute kidney injury; Hark, to manage clinical tasks and interventions. Here's Imperial College London's experience with Hark.
- Artificial intelligence startup MedyMatch launches, former Philips exec tapped as CEO
Current Weaknesses in the System
- Dependent on the accuracy of structured data
- Ability to interpret unstructured data still rapidly developing
- Disease is multifactorial - and treatment success is in part driven by patient compliance (human factor again)
No comments:
Post a Comment