Artificial intelligence in medicine and healthcare
What is AI anyway?

Artificial Intelligence (AI) is the name given to any form of technology that has been programmed to complete a task that a human would use intelligence to perform.

Let’s break that down a little.

We’d call it AI if the technology uses one or more of:

  • Planning
  • Learning
  • Reasoning
  • Problem solving
  • Knowledge representation
  • Perception
  • Motion
  • Manipulation

There have even been AI projects that seem to demonstrate social intelligence and creativity.

Why we love to hate AI

We’ve probably all been witness to the most intelligent of people get beaten down by self-checkout, but those same people who curse the machines that have replaced our checkout operators benefit hugely from the AI in the rest of their lives.

The generation gap that once saw our elders looking at technology in wonder is closing, with AI and tech in general creeping into every industry and sector in at least some way.

The rows of filing cabinets once common in many offices are now a thing of the past as the paper trail is replaced by computers.

We truly are embracing technology for all it can do, but it’s understandable that we’ll complain bitterly when we feel an immature technology is complicating our lives rather than helping us.

AI is in its infancy in many respects and while it helps almost all of us quietly in the background, people become acutely aware of it only when things go wrong.

Couple using self-checkout
AI everywhere

AI is in our pocket, it listens out for a cue word ready to answer our questions, sets us a reminder or even reorders our favourite washing powder without us having to move a muscle.

It’s undeniable that affordable tech is changing the way we live and manage our lives.

Every day it seems we introduce new smart tech to our lives that can reduce the amount of time we spend doing the stuff we don’t want to, so it’s natural to start to question what lies ahead of us in terms of AI.

The challenges facing healthcare

Healthcare is one of our more underfunded industries. No matter where you live, accessing affordable healthcare is becoming increasingly difficult. There are seldom enough doctors or nurses in any given area and the costs of accessing the ones we do have is growing.

What seems to make things worse is the fact that some medical conditions are so rare that the specialist skills required to diagnose and treat them may be absent from an entire continent, unless of course you’re fortunate enough to be able to self-fund and travel to the help you need.

This issue is set to only get worse thanks to an ageing global population, and treatment for conditions that were once just managed with compassionate care for the dying are now in higher demand than ever.

Life expectancy continues to soar. In 1900 you were expected to live until around the ripe old age of 31, while figures from 2014 had more than doubled to a more reasonable 71.5.

This is great news, except that having increasing life expectancy puts even more pressure on our healthcare systems.

Elderly person in hospital

Doctors and nurses are working themselves to the bone to be available to treat as many people as possible, but the waiting time for diagnosis seems to be longer than ever. Many people even in developed nations find that they’re having to wait longer than a month to see a primary care physician, with some people dying before they’re able to access healthcare.

Could AI transform healthcare?

While there’s plenty of concern about automation taking peoples’ jobs, could the introduction of this technology in support of medical professionals revolutionise healthcare?

Not everyone is keen to embrace new ways of using AI and simply suggesting that AI is about to revolutionise healthcare might send half the nation into panic mode. People don’t like to imagine a world where doctors have been replaced by robots.

However, no matter how smart our tech gets, even the more strident advocates of AI in healthcare don’t see a future run by robots. They see a partnership between humankind and machine.

At the end of the day, AI is only as intelligent as the data it has to work with. While AI can be taught to pick up patterns and anomalies based on algorithms and calculations, what AI may never achieve is a human level of compassion, understanding and empathy.

That said, try to imagine a centralised system holding the medical records of everybody in the world. When you consider how effective AI could become at diagnosing conditions there’s simply no argument for keeping AI out of our hospitals – provided it’s placed under careful supervision!

Why does AI scare us so much?

However, in a recent study, it was found that only 54% of those asked would be willing to ‘speak’ with an AI doctor about their symptoms and receive a diagnosis or advice.

What’s crazy about this is that many of these same people probably wouldn’t think twice about typing their symptoms into their favourite search engine, and then considering whatever they found to be gospel!

This is despite much of the content on the internet having been posted by people, like you and I, simply sharing their opinions and anecdotes, as opposed to an AI system which would be programmed and configured using medical records and senior doctor supervision.

What’s even more baffling is that most of us already use AI as part of our daily routine without realising it, yet still reject the idea of many of these advances.

We see people everywhere we go wearing Fitbits, Apple Watches or other smart devices that track our heart rate, temperature and movements to give us tailored advice about looking after our health. How far are we from syncing that to a national database to allow our doctors to have more insight on our lifestyle and habits, enable them to diagnose patients’ conditions more easily?

Doctor showing patient laptop
AI already delivering value in healthcare

Although these are not everyday devices, type 1 diabetics have long been offered the option to wear a pump attached intravenously to the their stomach or other suitable site. This AI device automatically tests blood sugar and provides the correct dosage of insulin without any need for monitoring, even notifying the patient when it’s time to retest, issuing a call to action when blood sugar falls.

What about pacemakers? A device that saves thousands of lives by looking out for changes in the pattern or routine a patient’s heartbeat.

Newer and more advanced types of AI are already been trialled in the healthcare industry. Most notably, robotic pharmacies or automated dispensing systems have been introduced at many hospitals. This technology could help to eliminate dosage errors and reduce the delays inpatients experience whilst waiting for discharge packs and medications.

Yes, this is going to affect pharmacy staff – but hopefully positively.

We rely more than ever on pharmacists to alleviate the strain on clinics and emergency centres, and pharmacy queues are often very long.

But pharmacists shouldn’t fear for their jobs – it will be a while before a robot is sufficiently advanced to replace a trained professional, who has the expertise and experience to deliver appropriate advice in a sensitive manner.

The tech to diagnose minor ailments such as rashes in children for example, would be so complex that it would take decades to for the technology to develop sufficiently and for it to be considered affordable.

But what if we could implement a system that worked alongside our pharmacist and dispensed drugs for them under their watch, without the need for as many checks or signatures before a drug can be released – how could that improve our system?

It’s important to look at what AI could save us with regards to resources, time and lives by just being there to assist our overworked team of healthcare providers.

Out-of-hours provision

It’s expensive to staff a pharmacy 24 hours a day. However, it’s much less expensive to leave a machine in situ so if a diabetic suddenly finds they’re out of insulin they can have their latest prescription pop out whenever they need it.

Or what about a patient discovering they’re without vital medications for their heart?

How many spaces in our A&E departments would be made available if prescriptions were digitised and available from any pick-up location?

Stethoscope alongside laptop
And what about training?

A groundbreaking study saw 1.3 million patients under 18 have their medical history, along with the symptoms and the reason for their visit to the doctor, logged into a system. The data consisted of both laboratory results and the notes made by doctors about the child’s presenting state.

After all the data had been input into the AI system it was found that the software could diagnose glandular fever, roseola, influenza, chicken pox and hand, foot and mouth disease with 90-97% accuracy.

Kang Zhang, who headed the research project stated “It’s not perfect, but neither are human doctors. When you’re busy you can see 80 patients a day, and you can only grasp so much information. That’s where we potentially as human physicians might make mistakes. AI doesn’t have to sleep; it has a large memory and doesn’t lose energy”.

In a doctor vs AI test, the AI out-performed junior paediatricians measurably. However, it did not quite match up to the senior consultants. Is this enough to make us consider using AI to assist our junior doctors in their training?

We use AI without even knowing it

What’s surprising is that this type of technology is already being integrated into our lives more than you realise.

There are many online ‘symptom checkers’ that depend on a number of factors to provide a flowchart question and answer format, ensuring patients access the right service according to their ailment.

Many A&E departments use ‘self-check-in’ kiosks where a patient checks in using a touchscreen interface. An algorithm is used to assign them to the correct team and with an appropriate priority score.

With this in mind; how far are we from the system ordering the tests that would enable your symptoms and clinical results to be available to your doctor before he even sees you, making for a diagnostic process which is more efficient than ever before?

Where minor ailments are concerned a doctor could simply review the results of these tests and send a prescription back to the kiosk for you to collect, without ever having seen you in person.

Lessening the impact of contagious disease

This type of diagnosis and treatment could offer further benefits by reducing the amount of time patients with contagious symptoms spend in healthcare facilities, reducing the amount of hospital-contracted illnesses for inpatients. This would also lower healthcare professionals’ exposure to viruses, reducing sickness rates.

Trainee doctors observing surgery
Still not convinced?

However, this is likely to be met with fierce resistance; whilst over a third of those asked thought that having AI in healthcare would make healthcare easier and quicker to access for more patients, around half felt that they would not trust an AI system to make an emergency clinical decision on their behalf.

But there are so many different applications for AI in a healthcare context, not just in terms of clinical or surgical intervention.

Our healthcare systems often run inefficiently or have high instances of sickness and absence, so how about we use AI to look at HR management?

AI can be used to pick up trends and issues with staff absence, meaning that if Sandra takes the last Friday of the month off with a migraine every month, it won’t take long for the system to spot that there may be something untoward going on.

AI could also help spot trends regarding lateness and attendance which may suggest personal issues or mental health difficulties, making it easier to provide information to staff to help them access the services they may find helpful.

AI has also found its way into accounts departments where systems are programmed to carry out constant low-level auditing of accounts and pick up accounting mistakes and cases of fraud, preventing money being lost due to employee mistakes or dishonesty.

This type of software could save healthcare system millions.

How to build trust in AI – start small

Using AI to replace human interactions will always be frowned upon by many, but we seem to selectively accept AI into our lives. Could our resistance be based on our lack of knowledge?

Asking someone a question as general as “Would you be happy to have AI manage more tasks within the healthcare system?” may illicit this sceptical response because people start thinking the worst, imagining a couple robots performing open heart surgery whilst the consultant takes a coffee break or goes out for dinner, when realistically we need to aim much smaller than that.

For example, a simple booth that could take a patient’s weight, height, blood pressure and pulse, and either approve or deny repeat prescriptions based on this data, could save our healthcare professionals time.

Add a touchscreen to ask questions about lifestyle, existing conditions and medications and you’ve just taken the pre-assessment nurse out of the office and put her on the ward where she was so desperately needed, improving the healthcare environment for all.

There is nothing stopping a doctor or nurse reviewing this information and bringing people in where the information provided is insufficient or where complications arise. But there are huge benefits to keeping simple, straightforward cases on an efficient path, allowing healthcare professionals to focus on the not-so-simple cases.

Although many of us would be unhappy with attending an appointment and coming out with a treatment plan without ever seeing a healthcare professional, there needs to be some trust in the system.

AI could eventually offer full audit trails and query professionals who don’t spot conditions that may be red-flagged in tests – potentially saving the system billions annually in medical negligence claims.

Robotic surgery
There’s enough backing to start phasing in AI

Whilst only 37% of us would allow an AI/robot to operate on us for invasive/major surgery, a massive 47% would gladly go under the robot’s knife for non-invasive/minor surgery. So while the major surgery patients are operated on by a surgeon, 47% of patients who opted for minor surgery by robots have freed up professionals to apply their expertise in more pressing areas. Were the provision of robotic surgeries to be correctly implemented, ever growing waiting lists could significantly improve over just a matter of months.

So there is mileage in having AI in our theatres, and – say it softly – it’s already there.

Many doctors are using AI systems that replicate their own movements. translating a cm of movement to a mm of robotic movement allows for incredible accuracy in the most difficult of surgeries. Technology exists that even detects and reduces the tremors of a human hand.

Speaking as someone who has been under the knife twice in as many years for spinal surgery, I wouldn’t hesitate to allow a robotic hand to perform the tasks the doctors did.

Some procedures will never be simple – they carry complex risks which can be nearly eradicated purely by introducing a robot that knows which bits it can’t touch. But in simpler cases or particularly where keyhole surgery is concerned, procedures are already performed with very little hands-on activity from surgeons.

We know we need our surgeons present, and more importantly, we want them there.

All surgery carries risks and should a patient suddenly suffer a complication during surgery, we want the surgeon there calling the shots, not an AI system with no pathway programmed to deal with the problem.

The future of healthcare technology

AI will undoubtedly play a significant role in the future of the provision of healthcare.

With the investment and start-up value for AI in healthcare set to shoot up by 50% a year to make it a $36.1bn industry by 2025, it may well be that trials and new innovations are made available that are so effective that we can only welcome them with open arms.

Personally, I’m eager to see medical advancement as a result of the development of AI and hope that by the time my children are grown, they’re able to benefit from a more efficient and accessible system than the one we have now.

What about you?


Related Posts

Leave a comment