Embracing AI in medicine
Regardless of whether you live under a rock or not, there is a strong chance AI will be a hot topic at the water cooler. The conversation will be somewhere on the spectrum of ‘I am excited’ to ‘I am afraid’. A large number of prominent names have spoken on the impending dangers of AI but I’m not looking to stir up the debate on a general AI taking up arms a la Skynet. Instead, this is a look at how AI will change our working lives, particularly where it will take work away but also where it will create new work.
One field in which we are seeing many in-depth conversations on the applications of AI is medicine. Cutting edge technology and medicine has always been a contentious topic as the stakes are higher than most other industries. Many see this as enough red tape to make it impossible but I feel there is huge potential here to make a large and beneficial change to healthcare without causing undue damage. There are those on the other side of the fence as there have been at many points in history but I think with due care and an open mind, medicine is in for a change like it has not seen in a long long time.
As a loss averse species the idea of something we’ve always had or done, suddenly disappearing is difficult. I’ve felt it in many facets from new software tools and libraries all the way down to debating the value of a food processor because ‘I have a knife, I can chop just fine’. While the contents of my kitchen cupboards is trivial, the prospect of losing one’s livelihood, not due to a personal failing, but simply because of being made obsolete is confronting.
When we put ourselves in the shoes of the organization, and in this case a hospital, as a whole there are two options. One — embrace the change. That means planning for an AI filled future. This option, gives an organization the best chance of survival. The other option is to fight or ignore the change. This option is sometimes the right one. Not every hyped up ‘disruption’ is real. There is however, a tipping point. A point at which the potential returns become too good to ignore. At this point resisting change is almost always to the detriment of the organisation.
This has happened countless times across many industries and doesn’t always come from radical shifts in technology. When the idea of hand washing was first proposed to clinicians by Ignaz Semmelweis in the mid 19th century he was ridiculed, committed to an asylum and died there. It was only years later that microbiology caused his ideas to become irrefutable and hand washing became widespread. This change eventually dropped obstetrical death from as high as 35% down to less than 1% of patients.
I shudder to think the same thing might happen again with the introduction of AI. Right now we have, without a doubt, some of the world’s best healthcare facilities. The problem however, is that the population growth is rapidly outpacing the rate at which we can train clinicians to cope with the demand for diagnosis, made worse by the number of conditions we can diagnose increasing as well.
If we refuse to shift some of the low hanging fruit to AI, we will have clinicians spending more of their time on things AI could do for them and less of their time focused on the work AI can’t do. I want to see a future where AI has every clinician working to push the boundaries of medicine, treat as many people as possible and make diseases like cancer a thing of the past.
To see this happen I think there are three things any of us, clinicians or not, can do right now to help this future come into reality.
Invest time and money in AI startups
As is often the case, innovation is founded in new ventures and then absorbed by the incumbents or live long enough to become them. The best fuel to the AI fire is using, interacting with and even talking about awesome AI focused startups. There are many AI powered startups in medicine right now and just like the pharmaceutical industry, there is room for many people working on deep problems. Here is a huge list from CB Insights showing what is probably only a portion of the startups applying AI to medicine.
Learn Machine Learning
With all the focus on teaching kids to code, adults need to know that right now, with no previous experience, there are hundreds of free resources to tech machine learning and AI. Whilst a few cowboys may go for AI for AIs sake I would hazard that picking a problem and throwing machine learning at it is a much better was to get started. Even if you fail to solve the original problem you’ll have much better context as to what AI can do and where is boundaries are. To get started here are some good, free, online resources.
- http://course.fast.ai/
- https://www.kaggle.com/
- https://www.udacity.com/course/deep-learning-nanodegree-foundation--nd101
- https://www.udacity.com/ai
- https://www.udacity.com/course/deep-learning--ud730
Listen to the worried
As I mentioned in the beginning there will be work that was previously done by humans that won’t need to be any more. If you run a company this could mean less people have jobs. Despite all our advances in computing I think it still remains true that people are the most catalyzing components of a company. Exploring the worries and making plans to supercharge your staff means jobs aren’t all being lost and you gain more than just the time saved by replacing a human with a machine. Sadly for this I don’t have any good links. So I’ll leave that as ask for the reader. Make this happen and send me links. I want to hear these discussions, in medicine especially as the only way we can make this happen is to combine the knowledge of engineers, clinicians and all other people in the complex system that is healthcare. I’m always more than happy to discuss these ideas online or in person. Twitter is probably the best place to reach out if you want to discuss. I am happy to hear opinions of both sides of the fence.
