Image Credit: sdecoret/shutterstock.com
By Chloe Bennett, BSc
Globally, health services are facing a plethora of challenges caused by greater demand for services, an aging population and health expenditures. Further strain can often be placed on providers due to poor productivity and efficiency. Artificial intelligence is argued by some to be pivotal in the transformation of health providers such as the NHS as its promised to “achieve more by doing less”.
What is Artificial Technology?
Artificial intelligence (AI) is an umbrella term commonly used to define the type of computer science which involves the simulation of human cognition in computers. The use of AI has grown immensely since its conception in 1956, and now, has many applications extending from its use in robotics to personal assistants such as Siri and Alexa, to healthcare services such as the NHS.
There are two main subsets of AI; machine learning and deep learning. In machine learning, after being trained to recognize patterns, learn rule-based logic and the different types of reinforcement strategies, software algorithms aim to predict the occurrence of future events. Deep learning is a type of machine learning which uses a series of artificial networks, similar to synapses and neurons within the human brain. Each network is comprised of a range of interacting layers which all perform different tasks.
There is a growing body of research which highlights how artificial technology can enhance healthcare services such as the NHS.
Improved Test Readings
The use of OsteoDetect, a detection and diagnostic tool which relies on machine learning, was approved for use in 2018 by the United States’ Food and Drug Administration (FDA). OsteoDetect aims to detect potential wrist fractures on film X-rays and was devised to be used alongside Radiologist’s observations. Research investigating the effectiveness of the AI found that it enhanced Radiologist’s specificity and sensitivity values when compared to diagnoses made without AI assistance. Ensuring that diagnoses given to patients are accurate is extremely important. It is thought that diagnostic errors can pose a threat to safety and quality as the errors made could have harmful effects for the patient. If similar technology were adopted in the NHS, particularly in busy Accident and Emergency wards, it could improve the accuracy in care received by patients, and in turn, reduce waiting times and demand on hospital services.
Reduced Waiting Times for Referrals
Research into the use of AI in healthcare settings has shown that software programs may aid routine duties. The FDA also recently approved the use of IDx-DR; a cloud-based AI program. With the use of the technology, medical professions were able to use a retinal camera within their office to take images of patients’ retinas for screening purposes. These images could then be uploaded for analysis by IDx-DR. Timely feedback would be given noting whether the images suggested the presence of conditions which would require a referral to an Ophthalmologist.
By speeding up detection of medical conditions, the use of AI could help reduce waiting times for referrals. This could have further benefits as early detection and treatment could lead to better management of health conditions.
Reduction in Unnecessary Treatment and Diagnosis
There is a growing problem in healthcare; overdiagnosis of conditions due to increased sensitivity in detection methods which is resulting in overuse of healthcare. In the USA, it is thought that approximately a third of those diagnosed with cancer may reflect an overdiagnosis and that 30% of those diagnosed as being asthmatic may not have the condition at all. If similar errors are experienced in the UK, then AI in the NHS could be used in healthcare management systems to analyze any unnecessary diagnoses and treatments and reduce expenditure.
Early Detection and Prediction of Conditions
There is a growing body of literature outlining the positive implications of using AI to assist in the diagnosis of medical conditions. Researchers have demonstrated that the use of deep learning AI was slightly more successful compared to Dermatologists in differentiating between melanomas and benign tumor moles on imaging. Further research has found that upon evaluation of brain PET scans, learning software was able to predict which patients presenting with cognitive impairments were more likely to develop Alzheimer's in the following two years.
Similar research involved the use of deep learning AI to study over 950 lab and clinical test variables per patients to predict the onset of health issues. They found that the AI was able to predict the development of diabetes more accurately compared to previous prediction models. Use of such technology in the NHS could significantly improve the quality of life of many patients through early detection, which could allow for better management of symptoms. It could potentially reduce demand on resources if diagnoses are given in a more timely manner, as exploratory testing of symptoms presented may not be required.
Improved Patient Assistance
AI can be used in assistive medical devices and robots. Applications include walking devices which can be used to assist with sitting, maneuvering, standing or walking; and telerobots, which can assist with communication pathways between patients and medical professionals. If such devices were used in the NHS, it could help to free up the time of medical professionals in hospitals and improve patient care.
Despite a range of potential benefits, attitudes towards AI may need addressing before patients can accept its use in the NHS. Recent surveys in the UK, report that over 60% of respondents are uncomfortable with their personal information being used to improve health care provision and are unfavorable to AI being used in replacement of nurses and doctors.
Sources and Further Reading
- Guo J., & Li B. (2018). The Application of Medical Artificial Intelligence Technology in Rural Areas of Developing Countries. Health Equity. DOI:10.1089/heq.2018.0037
- Panch T., Szolovits P., & Atun R. (2018). Artificial intelligence, machine learning and health systems. Journal of Global Health. DOI: 10.7189/jogh.08.020303
- Vayena E., Blasimme A., & Cohen G. (2018). Machine learning in medicine: Addressing ethical challenges. PLOS. DOI: 10.1371/journal.pmed.1002689
- Craft J. A. (2018). Artificial Intelligence and the Softer Side of Medicine. The Journal of the Missouri State Medical Association. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6205273/