Artificial intelligence is an ubiquitous entity of the 21st century, replacing almost every process that is repetitive and protocol-driven. The effects of artificial intelligence have been witnessed in a spectrum of areas from home-cleaning personal assistants to machines involved in microsurgeries.
A group of researchers at the Massachusetts Institute of Technology, USA (MIT) devised a form of artificial intelligence that can identify relationships between variables. Artificial intelligence with the ability to identify relationships is vital. This ability is valuable in medicine which is propelled by the modification of risk factors and disease prevention.
In the 1980s, Judea Pearl revolutionized artificial intelligence by inventing machines that could attribute specific outcomes to a possible causative variable. For instance, for an individual with fever, muscle aches and a history of recent travel to Africa, a possible diagnosis of malaria could be postulated. This association identification is vital as it simplifies the complex process of diagnostic reasoning that governs the practice of medicine.
If artificial intelligence in the current decade is equipped to provide a rationale for causative links, several medical conditions can be prevented. Recognizing the link between alcohol consumption and chronic liver disease is an example of an association that machines can be trained to identify. Having identified the possible root cause of a condition, it is possible to delve into risk modification for that particular condition.
Whilst risk factors of many medical conditions are well established (for example, the link between smoking and heart disease), there are several medical conditions for which causation remains unclear. One example of this is the link between chronic lung disease and no history of smoking.
Babylon Health, a digital healthcare service based in the United Kingdom, has taken this feature of artificial intelligence to identify relationships one step further. Researchers at Babylon Health have devised a virtual doctor – an app that identifies symptoms (e.g. chest pain, cough), creates the most plausible diagnosis and provides advice about management strategies. This form of healthcare provision can be very cost-efficient as it saves both the provider’s and patient’s time. It also allows individuals to quickly obtain a diagnosis and find out what is causing their symptoms, as opposed to being put on waiting lists for a consultation which can be a lengthy process.
Randomised controlled trials investigating drug efficacy or possible correlation between controlled conditions are often time-consuming and expensive. These trials can involve considerable human resource investment and time to interpret results. Artificial intelligence that is trained to spot links between cause and effect based on a wealth of statistical data already present can enable quicker identification of factors linked to disease. For example, the role of the Ebstein-Barr virus (a virus which causes glandular fever) in the causation of multiple sclerosis has been debated for several years. Artificial intelligence that collates data which supports or refutes this hypothesis can save a lot of time that would be spent on research and data collection.
To truly propel artificial intelligence forward, more sophisticated techniques to generate solutions is required. For instance, if a low-fibre diet is linked to an increased risk of bowel cancer, we need a way to assess the effect of widespread interventions to increase fibre intake on the general risk of bowel cancer. This elevates the simple process of cause and effect identification to a pragmatic, applicable solution which has the ability to benefit society.
This form of technology is particularly useful for today’s generation. Today, people take a keen interest in their health and are constantly looking for ways to optimise their health and well-being. An app at the tip of your fingertips that can tell you if you are at an increased risk of certain diseases can help you gain greater control over your own health.
Knight, Will. “If AI’s So Smart, Why Can’t It Grasp Cause and Effect?” Wired, Conde Nast, 9 Mar. 2020, www.wired.com/story/ai-smart-cant-grasp-cause-effect/.
Hartnett, Kevin, and Quanta Magazine. “To Build Truly Intelligent Machines, Teach Them Cause and Effect.” Quanta Magazine, www.quantamagazine.org/to-build-truly-intelligent-machines-teach-them-cause-and-effect-20180515/.
Heaven, Will Douglas. “An Algorithm That Can Spot Cause and Effect Could Supercharge Medical AI.” MIT Technology Review, MIT Technology Review, 5 Feb. 2020, www.technologyreview.com/s/615141/an-algorithm-that-can-spot-cause-and-effect-could-supercharge-medical-ai/.