Harnessing AI for healthcare

October 15, 2018 0 By CH Unnikrishnan

Dr Arun Bhatt

I believe this artificial intelligence is going to be our partner. If we misuse it, it will be a risk. If we use it right, it can be our partner.” – Masayoshi Son, Japanese business magnate.

Artificial intelligence (AI) is the name given to techniques that use software to mimic human cognition in the investigation of complex health data and information. AI is revolutionizing healthcare with its access towards a large amount of health data and rapid advances in analytical techniques. AI is considered augmented intelligence as it can help in reducing diagnostic and therapeutic errors and provide real-time interpretations for health risk signals and health outcome prediction. AI systems analyse a variety of data – clinical examination, laboratory data, diagnostic imaging, genetic testing and electrodiagnosis – to improve medical decisions.

AI devices consist of 1) machine learning (ML) techniques and 2) natural language processing (NLP) methods. ML procedures analyse structured medical data – e.g., imaging, genetic and electrophysiology, to cluster patients’ characteristics or to determine the probability of disease outcomes. NLP methods evaluate information from unstructured data – e.g., clinical case notes and medical journals to complement and enrich structured health data.

AI applications have the potential to be used in diverse health conditions. However, most of the AI research focuses on disease areas with high mortality – e.g., cancer, nervous system disease or cardiovascular disease. AI systems have been used for diagnosis of skin cancer from clinical images, diagnosis of cardiac disease using cardiac images, to restore the control of movements in patients with quadriplegia, early detection of Alzheimer’s disease, early stroke prediction by employing a movement-detecting device and predicting and analysing the performance of stroke treatment. AI has also been utilized to detect congenital cataract disease using ocular image data and diagnose diabetic retinopathy through retinal fundus photographs. The accuracy, specificity and sensitivity of AI diagnostic systems are high and superior to those of experienced physicians. However, AI is still in the early stages of research and adoption in medical care. Besides, its real-life implementation is facing regulatory challenges.

AI pose a new challenge for physicians, some of whom are concerned that this advanced technology will replace them. But the fears seem unfounded. AI is unlikely to replace the doctors but is expected to optimize and improve their medical skills. Dr Abraham Verghese, a well-known author from Stanford University, opined that “the two cultures – computer and physician – must work together (JAMA Jan 2, 2018).” AI has the potential to reduce tedious administrative tasks, handle large volumes of medical
data and information, and provides an opportunity to integrate the expertise of medical practitioners and intelligent automation.

AI is called the stethoscope of the 21st century, says Dr Bertalan Meskó. The tech-savvy physician of the 21stcentury should not fear AI tools, but should be ready to harness benefits of augmented intelligence in her practice.