To date awareness of the current and growing capabilities of AI has barely penetrated the busy world of Healthcare. As such the opportunities to improve both the effectiveness and productivity of the Healthcare system remain visible to very few – despite a number of commentators publicly stating that the Return on Investment of AI in Health could be greater than any other industry.
Some of this recognition issue is not specific to Health. The rapid improvement in speech recognition in recent years has been enabled by improved understanding and use of LSTM deep neural networks. A recent publication by Microsoft has demonstrated that, for conversational speech, recognition is now at or near human levels. In addition, the success of Amazon’s Echo products demonstrate that AI-based speech recognition is now ready for everyday use, and with ongoing improvements will work in more specialised arenas. This AI enabled technology will likely revolutionise many workplaces – including Health – and yet for most the technology remains either unknown, or considered of passing leisure time interest only.
Within Health there are growing examples of the power and utility that AI can bring to improve patient care. The recent work by the Standford team, using Google’s Tensorflow to achieve opthalmologist-level recognition and grading of 130,000 diabetic retinopathy images, is system-changing for many parts of the world. It will allow earlier recognition and thus management of one of the fastest growing causes of blindness in the world, while freeing opthalmologists from a repetitive screening task to concentrate on more challenging therapeutic endeavours. Similar results from the same institution have been achieved for skin cancer recognition – a disease with better than 95% survival if recognised early. The plan is for this technology to be available on smartphones, with the potential to save thousands of lives.
The key to changing this is for all people involved in AI to regularly discuss success cases resulting from AI application, both to colleagues and on blogs. This will then slowly move the discussion away from fear towards the increasingly positive benefits that AI is bringing. Eventually AI will be accepted as simply business-as-usual.