Introducing AI in Pharmacy
During the past few decades, AI has evolved so much that it has become an extremely valuable tool for improving almost every aspect of life: education, customer service, cybersecurity, engineering and planning, economy. Healthcare is no exception, but, even if less obviously, pharmacy as well.
But, how did this happen? Pharmacy is a notoriously conservative field, and it has existed for thousands of years. AI, on the other hand, is still relatively new and, at least at first glance, seems like the exact type of imperfect technology pharmacists should more likely avoid than embrace.
Yet, as they have already realized, there are plenty of ways AI can help them. In this article, we look over merely few of them.
AI and Predicting: Better Than Humans
The human brain is an inscrutable wonder of nature, but it isn’t exactly efficient when it comes to dealing with huge volumes of information. This is where AI comes in handy: it’s millions of times more proficient at gathering and analysing large datasets than human intelligence.
We, as humans, are not really designed to remember details, even if given the chance to scrutinize something for hours and asked about a certain detail only moments later. If you don’t believe this, just try to read a book of an average length and remember what a character was wearing on page 57. But, what if this was the most important symbol in the book or the answer to one of the questions on your final exam?
The point is fairly simple: we can easily overlook important or even crucial information and, consequently, completely miss the whole picture. AI programs, on the other hand, can efficiently recognize patterns in thousands and thousands of seemingly unrelated cases and create predictions based on these patterns. This is as impressive as it sounds: when it comes to large datasets, AI will always notice more than us in an incomparably shorter period of time.
And what about the subsequent predictions? Well, this is where it gets interesting.
Predicting the price of a new house based on an input/output dataset comprised of houses whose prices are already known is a pretty straightforward example. But predicting the risks a defendant may present to society based on his previous behaviour is something little less obvious. So is predicting the future prospects of your employees based on their ambitions or previous work experience. Legal institutions and large corporations, however, have already installed AI programs designed to make these predictions exactly.
And if law can profit from AI, why shouldn’t pharmacy as well? In the pharmaceutical industry, analogously to the examples above, AI can be used to predict the outcomes of using certain pharmaceutical by patients with different conditions. The main reason is that biochemicals usually contain a lot of constituents each of which can have varying impact depending on the patient and his medical history. Human brains are incapable of keeping them all in mind, but computer programs, this is merely a question of processing power. And since quantum computers are already on their way, soon enough, it won’t even be about processing power anymore.
Real Applications of AI in Pharmacy
There’s an obvious interest among many Internet users in the possibilities of being replaced by robots. That’s neither new nor strange. In fact, many jobs may become obsolete once AI is properly implemented and this may even lead to putting into practice some interesting new economic strategies, such as UBI (read the section on “Job opportunities” here for more). What about pharmacists, though? Is there any risk for them to be replaced by robots?
According to an exhaustive study authored by two Oxford scholars and titled “The Future of Employment”, there’s a comparatively low risk for pharmacists to be replaced by robots: just 1,2%. This puts their occupation right at the bottom of computerisable occupations, lower than, say, craft artists, writers or social workers (54th out of 702 analysed professions, 1 being the least probable job to be computerised, in this case recreational therapists). However, the fact that pharmacists are so far safe from being replaced by robots doesn’t mean that they shouldn’t be interested in being helped by AI either. On the contrary, in fact: AI can already be broadly applied in the field.
Usually, we tend to see the work of a pharmacist in terms of a simple sequence of actions: he takes the prescriptions from the doctor and gives the appropriate pharmaceuticals in response. However, this is only a part of pharmacist’s actual work which includes broad knowledge of biochemistry and physiology. And when knowledge constitutes a part of the equation, technology is always helpful.
Helping patients to take care of their health is the most sacred of all pharmacists’ duties. This implies assisting patients in developing the right kind of attitude to the prescribed medicines and understanding the tools for measuring some functions of the human organism. As new devices and pharmaceuticals are rapidly entering healthcare, this task is becoming ever more relevant and important.
However, smart devices using AI technologies can tell pharmacists a lot about a patient’s condition in real-time. Pharmacists won’t need to rely merely on subjective opinions anymore – such as how a patient feels or how he or she looks. Instead, it should be possible in the very foreseeable future to instantly figure out the reaction of patient’s organism to the prescribed pharmaceuticals (by reading its effects from these devices).
There is already a bunch of smart devices on the market – such as Nokia BPM+, Muse Headband, AliveCor, and many others – which detect what is happening in a person’s organism in real-time, something incredibly useful for patients, doctors, and pharmacists. The data acquired from such devices can be used later in Artificial Neural Networks (ANNs) to predict potential developments of diseases and, thus, help in preventing them.
Wherever there’s data, there’s a possibility for perfecting AI and, fortunately, we have amassed a lot of data and developed different ways to use and interpret it. A lot of things are already happening in healthcare: just think of projects such as Sensely or BabylonHealth, or, of course, Google DeepMind and IBM Watson for Oncology (if you have somehow missed them). Using large piles of patients’ data, these tools help to build predictive models for diagnosing, preventing diseases, and treating patients.
In the very near future, pharmacy will be able to absorb all these innovations (perhaps, even earlier than many other fields), which may in turn open many more opportunities for pharmacists to apply their skills and knowledge to further develop the field.
Is There Any Place for Failure?
Unfortunately – as much as obviously – there is some. Despite being touted as smart ones (and despite outperforming humans), neither of these programs works perfectly. And the worst thing is that more often than not we don’t really understand why exactly they have failed.
In her book Weapons of Math Destruction, data scientist Cathy O’Neil describes various cases where machine learning algorithms (on which AI is based) have failed and failed terribly, causing serious problems for many innocent people. She calls such algorithms destructive and explains why so through the following example in an interesting interview she gave last year:
“Michelle Rhee in Washington, DC, was this really gung-ho education reformer. She was hired to apply all of these new reform ideas. She instituted both a bonus for people who got good teacher assessment scores, and for principals at those schools, and she would fire teachers with bad enough scores. What happened, we have reason to believe, is that in DC a lot of teachers just cheated. They like, changed the answers on their students’ scores.
It’s obvious to everyone that if you incentivize something like good standardized test scores, then the teachers are going to teach to the test. But it should also be obvious that if there’s enough carrot and/or enough stick, it’s going to be more extreme than that. You’re going to actually see cheating. And that’s what we saw. We saw a dubious and unusual number of erasures at various schools”.
Apparently, these kinds of things can happen – and are happening – in every field of human endeavour. That’s why Google updates its bots’ strategies constantly to prevent cheaters from getting better website rankings through cheating. With pharmacy, however, it’s much more serious than updating once a mistake has been made. The motto is much better formulated as “perfected before being released”. The closer we are to AI being implemented in healthcare and pharmacy, the more intensively and carefully it needs to be tested. That’s why scientists are constantly trying to make their algorithms more and more transparent so that they are able to understand what went wrong – if or when something does. That’s why new laws adapted by the EU and the UK intend to force this into being the norm.
All in all, it’s still way too early to give AI programs control over the human bodies. In fact, the more advanced an algorithm gets, the more challengeable and testable it becomes. That’s why the tools we mentioned above garnered a lot of attention. That’s how we know that each of them still has a long way to go before it comes into actual and everyday pharmaceutical and healthcare practice.
As we collect more and more data and develop more and more tools for interpreting it, it seems that every area of human life starts actively embracing smart technologies. Slowly but surely, pharmacy is becoming one of them. Information gathered from patients has been used for treating and preventing diseases since time immemorial, but, in theory, AI should be capable of using this data more efficiently and properly, by extracting patterns and making predictions based on these patterns. Being a pharmacist is a responsible commitment and pharmacists can be sure that this guarantees that their occupation will not be computerised in the foreseeable future. Consequently, they should start learning how to profit from the new AI-based devices which may make their profession even more significant as a sort of an important link between the patients and the doctors of the future.