In recent years, Artificial Intelligence has been applied to a vast number of new fields, with an ever-expanding list of possible new uses. One of these fields is Medicine where, as Maksim Richards explains, AI has the potential to revolutionise how we diagnose and treat disease.
As you walk down the drab corridors of a hospital, it can feel impersonal and soulless. However, you have healthcare workers who smile at you; talk to you and set you at ease while they treat you. Take these people away and replace their voices with mechanical whirring and you have a doctor who takes no interest in you and has no capacity to comfort you because it doesn’t understand your concerns or fears. This idea of automated medicine, powered by Artificial Intelligence (AI), is thankfully unlikely to happen. But today, AI is shaping the modern way of practising medicine and aiding human doctors in their decisions. Working together, AI and doctors can make better diagnoses quicker whilst developing medicines and treatments at an unprecedented rate and overall improving global health.
Artificial intelligence is “a system that perceives its environment and takes actions that maximise its chances of achieving its goals”. That is to say, AI is a computer program that writes its own code and can change the way it acts without human intervention. In medicine, when AI is mentioned, it is often machine learning where a computer is given a huge amount of data from which the AI can identify patterns, make decisions and even predict what might occur in the future.
AI can be applied in a seemingly infinite number of medical scenarios, but an example that has seen significant benefit is in the field of sepsis. Sepsis can occur when the body is fighting an infection. It is a hyperactivation of the immune system which is often fatal. Around 48,000 people die from sepsis in the UK each year representing around 1 in 5 of all deaths. Sepsis is treatable, only if caught early but unfortunately, it is very difficult to do so and even world-leading academics cannot agree when diagnosing it. This is where the Targeted Real-time Early Warning System (TREWS) comes in. This is a machine learning tool that used the health records of tens of thousands of patients to learn what patterns of symptoms, interventions and test results increased the risk of developing sepsis. Following its training, TREWS was then allowed to analyse the health records of new patients to give them a score based on how likely they would go on to develop sepsis. The results were staggering: doctors assisted by TREWS diagnosed sepsis, with a 60% increase in accuracy and 24 hours before their condition became life-threatening. This increase in sepsis detection with more time for the physician to treat could have a world-changing effect on such a burdensome illness.
AI’s raw power cannot be exemplified better than in drug discovery where it can analyse data that no human could hope to – even if given all the time in the world. For coronavirus AIs around the world were set on data banks of billions of different molecules and approved drugs and simulate lab tests to screen for ones that might act as antivirals to fend off covid-19 and bring death and hospitalisation rates down. This simulation speeds up the drug developing process but also offsets the cost of carrying out the manual testing of millions of compounds that wouldn’t work and focusses researchers on just the few that might stand a shot at fighting this pandemic.
Diagnosis and treatment require that patients are monitored either in the hospital or in regular checkups where often they forget the answers to questions about their health. The advent of wearable technology has begun to change all that and the apple watch is notable amongst these as the most worn watch in the world. These tiny computers can monitor your heart rate and even perform medical-grade tests like an ECG to look at how your heart is working potentially removing the need for prolonged hospital stays. For those who have suffered heart attacks, this is a fantastic tool for keeping an eye on their heart and can enable the data to be given to your doctor so they can make decisions about your care. But this is not where apple stopped and with the use of AI, they have begun diagnosing atrial fibrillation (AF) from a patient’s wrist. AF is an irregular beating of the atria of the heart that can lead to strokes, heart attacks and more fatal heart rhythms. AF is a common condition that is increasing in prevalence and if left untreated, can lead to strokes and heart failure. Using machine learning, the apple watch can detect atrial fibrillation from the ECG and notify the user that they might have a heart problem that should be checked (and confirmed) by a doctor. Usually, AF is diagnosed only after symptoms emerge but with these tools, it can be detected long before to prevent any damaging effects. With the tens of millions wearing these watches the largest database of ECG data could be created and if we were to combine this data with complete health records and genetic information, AI could reveal patterns in treatments and diagnoses that could revolutionise the way we manage these patients.
The future of healthcare may lie in AI to make diagnosis quicker, treatment more effective and overall, healthcare better. AI is bringing about change in medicine quicker than ever before and this is only accelerating, but the days of Dr Robot are far off and for now, these programs act as tools for the modern physician.
Maksim Richards is a Medicine student at St John’s College, Oxford
- Artificial Intelligence in Medicine: Applications, implications, and limitations – Science in the News
- Artificial Intelligence in Medicine
- Early-Warning Algorithm Targeting Sepsis Deployed at Johns Hopkins
You might want to explore our courses in: