Artificial Intelligence in Medicine

Artificial intelligence may be the future of healthcare, but exactly what does this mean for healthcare professionals and patients?

July 2021
Dhillon Hirani (Blogger)
Queen Mary UoL - 4th Year Medical Student

What is Artificial Intelligence (AI)? 

Artificial intelligence is becoming increasingly popular in everyday life without us even realising. It is a field of computer science which involving using a program to perform a task that might otherwise require a human, effectively simulating human intelligence or bettering it. Several things that we use and take for granted everyday have artificial intelligence at its heart - think Google/Apple Maps compared to using a physical map or autocorrect instead of opening a dictionary to check spellings, artificial intelligence is all around us and it is only going to get bigger and better. The value of the AI industry is predicted to go from $4 billion in 2016 to $169 billion by 2025.

Why is Artificial Intelligence important in medicine?

Artificial intelligence in general has several potential advantages compared to a human doing the same job. In simple terms, it can be quicker, more accurate, solve complex problems and can potentially work 24/7. This makes artificial intelligence almost invaluable in any task that involves repetition thereby improving healthcare efficiency and reducing human error; however, it isn’t without its faults.

How is AI used in healthcare?

AI in medicine is commonly split into two types, virtual and physical. Virtual AI deals with computer-based work such as electronic record systems or decision making whereas physical AI can include robot assisted-surgery or smart prosthetic devices.

Electronic health records are a broad database where patient information is stored and accessed by clinicians and sometimes patients. Online health records have become a norm in society and is heavily influenced by artificial intelligence already with features like online appointment scheduling, reminder notifications for immunisations and warnings of drug interactions amongst others.

The use of AI so far in medicine has been in automating tedious tasks as discussed above but the really exciting part of AI is its role in diagnosing diseases as well as its physical AI applications. Within diagnostic medicine, there are several studies taking place trialling AI software to work alongside doctors to analyse images/data. These systems are coded to learn patterns from thousands of previous images and are currently being developed to analyse chest radiographs, biopsy tissue samples for metastatic breast cancers, electrocardiograms/ECGs, and even diagnose mental health conditions through speech and facial patterns. For example, this means the system is able to recognise what a lung infection or tumour looks like on an x-ray or scan, and can therefore assist doctors in diagnosing and deciding the next step. A great example of the progression of the use of AI in diagnosis is DeepMind, a software developed by Google to diagnose eye conditions from retinal scans. Imaging analysis through AI can lead to more specific and accurate diagnoses; so far several studies have shown it to generate similar detection rates to experienced radiologists/pathologists.

However, imaging goes beyond just diagnosing, and likewise this technology can be extended to monitoring conditions over time such as during cancer therapy. With imaging activity increasing in the NHS, using AI can help reduce the burden on radiologists and make them more efficient at their job, not through replacing them but by working alongside them. AI also has a role to play in personalised medicine as it becomes more advanced and can calculate how specific pathologies are best treated in particular groups of people to increase the effectiveness of how we manage individual patients.

Other examples of AI in medicine that you may be more familiar with is the use of smart wearable devices such as the Apple watch and Fitbits. Beyond step counting, sleep monitoring and notifications, watches are becoming more health inclined with Apple Watches now able to take a rudimentary ECG that can detect atrial fibrillation, a type of rhythm disorder of the heart. This information can then be shared with a healthcare professional so the patient can get checked further to receive an earlier diagnosis. Other wearables include The Elvie Pump which is a discrete breast pump that can fit into women’s bra; real-time blood glucose monitors which can be used to control insulin delivery and lastly sensors on the body that can assess walking patterns and posture for neural disorders.

How is AI used in surgery

One of the most significant uses of AI is its role in surgery. Robotic surgery is becoming more popular (and is another broad topic in itself) and involves a degree of AI. The most popular device is the da Vinci robot which is used in several keyhole surgeries. In robotic surgery or robot-assisted surgery, a surgeon sits at a console in the corner of a room whilst the robot is at the patient’s bedside. With an almost video-game like experience, the surgeon is able to control the arms and tools of the robot from the other side of the room (potentially even the other side of the world via 5G) with incredible precision led by artificial intelligence. Robotic surgery has several advantages such as improved outcomes and quicker recovery times and is starting to be used more in several specialties.

As well as carrying out surgery, AI also has a role in training and planning surgery. Surgery simulation technology has become increasingly more realistic in the same way video games have with the introduction of virtual reality. Surgical simulation can be used to train surgeons by emulating carrying out a procedure in an operating theatre, so that they can repeatedly practice their skills in a safe environment with no risk to patients. As well as medical education, simulation can be used for planning complex surgeries where technology can mimic the anatomy of the patient so that surgeons can foresee the potential complications that might occur.

What is the future of AI in medicine?

AI is moving fast and is being continuously built upon and expanded. Its role in patient databases will be built upon to do even more, robotic surgery and simulation surgery is becoming increasingly more realistic and accurate, and using AI for diagnosis will be further developed to combine various data sources from imaging, blood tests, and even genetics to improve diagnosis and management of patients. The applications are endless, but the real advances will occur when we are able to push AI use within developing countries where human expertise may be sparse, such as using AI to analyse radiographs and diagnose TB in countries with high prevalence but few trained radiologists.

What are the risks of AI in healthcare?

AI is a growing field and in general there is a lot of worry surrounding the use of AI in society with predictions about how common jobs in customer services and banking may be completely automated in the near future. People are naturally anxious about whether they will lose their jobs to machines in future, and what impact this will have on society. However, within healthcare, it's important to remember that medicine is patient-centred and ultimately the management of a patient isn’t as simple as a treatment for a disease. It takes into account a multitude of factors to ensure absolutely everything is being done in the patient’s best interest. This requires a degree of human intuition to create a trusting patient-doctor relationship that cannot be recreated in a computer program. It is only through human traits, such as empathy, communication and active listening that we can ensure patients are managed holistically.

Creating these AI programs are also very time-intensive and expensive, they can take years to develop and millions of pounds to do so with no guarantee to produce a working, reliable system. Robots do not have emotions, they can’t recreate a human interaction, nor can they think beyond their means which means any system that is developed will always be limited by its coding. Therefore patterns on a scan that may be a bit unusual for example, might not be picked up by the AI system, unlike a human who can use lateral and out of box thinking to solve the problem.


Another common issue is what is known as the block box problem. In AI, Information or data is inputted into an algorithm to produce a response. However, we don’t know exactly how or why it has given that answer, which makes it difficult to monitor the decision pathway to troubleshoot errors. On the other hand, you can always ask a clinician how they have come to a particular diagnosis, and they will be able to explain all the different factors that led them to their conclusion.


AI requires a lot of data, and whenever significant data is being handled there are always ethical questions to be asked of both the input and output data. For example, should patient data and records be readily available to organisations to create better AI systems? Or should the output AI data like information from an Apple Watch be available for healthcare systems or insurance companies to access? These barriers of ethics and trust in an AI system can limit its progress.

Why should we use AI in healthcare?

Overall, AI has great potential in healthcare with its benefits of speed, accuracy and potentially reliability. Although there have previously been question marks over whether it will ultimately replace humans in healthcare it is becoming increasingly clear that despite AI having a bigger role in medicine in the future it will be supporting healthcare workers, not replacing them.

This blog has just touched the surface of artificial intelligence in medicine, to learn more about AI in medicine for FREE:

Author: Dhillon Hirani

Editor: Latifa Haque

Suceed with the FREE Medmentor platform.