This post was written by Brendan DUNPHY, CEO of C-BIA Consulting and first published on AIMed
The digital world is not a new one, its just been slow arriving in healthcare. ‘What can be digitised, will be digitised’ is an old mantra. Atoms are converted into bits because information proxies of reality are cheaper to store and move. Once digitised, we can do a lot more with them than we can with atoms. This is probably truer in healthcare than anywhere else where well-defined ethics constrain research and limit interventions. By building digital models of patients, conditions, treatments and even clinicians, we can safely experiment and learn and then apply what we have learnt to patients in the real world, lowering risk and improving outcomes for all parties. ML and AI plays a major role here as it can see through the complexity and reveal insights we cannot.
Doctors get better with experience, so a challenge is how to ‘accelerate experience’ and thereby reduce error rates. One solution is virtual patients to expose trainees to a wider range of presenting patients and conditions, another is virtual doctors.
Its worth remembering that the enabling data analytics processing and storage technologies continue to increase in speed and reduce in price at a fast pace, especially GPU’s, making cloud-based solutions ever faster, cheaper and more powerful, crunching volumes of data inconceivable even a few years ago and delivering results in seconds rather than minutes or hours.
The scope to improve patient outcomes, lower risk and improve efficiency in healthcare by integrating data into clinical decision making is significant.
Current medical imaging can be very ineffective, especially in the UK which has one of the lowest number of radiologists per head of population in Europe so also means delayed prognosis. For example, 20% of lung polyps can be missed. The good news is that data standards are well developed and applied in the medical imaging domain making technical interchange of data possible. The case for ‘deeper tech’ in imaging and diagnostic areas to consistently go beyond existing human capabilities is easy to make and the trend is clear. More and more applications outperform even the best clinicians, and some are not simply ‘black boxes’ incapable of explaining their results. Next generation systems from imaging providers such as Siemens will go further, using not just image analysis but raw imaging data and data from other sources (including genomics and patient profile data) to present a more sophisticated and integrated diagnosis and one that may reduce the need for biopsies.
The UK’s National Health Service (NHS) is far from a single cohesive entity as often presented and perceived in the public psyche. It is a sprawling federation of 15,000 loosely connected public and private entities struggling to stay focused and aligned. There are probably 800+ IT organisations, none of which today could be considered on top of their challenges, or equipped to be so, let alone making best use of data or developing AI or ML to allow the NHS to escape from its past.
One promising area is diagnostic tools where software is already used to support Intensivists reach a prognosis.
This is a review of AIMed Europe 2018, learn more about the leading global event for clinicians in AI in healthcare here.