AI tool twice as likely to diagnose heart condition early, says US study

Dr David Rushlow from the Department of Family Medicine, Mayo Clinic, Rochester, USA and the lead researcher of study, said the team’s aim was to compare the clinicians’ characteristics of “high adopters” and “low adopters” of an artificial intelligence (AI)–enabled electrocardiogram (ECG) algorithm that alerted for possible low left ventricular ejection fraction (EF) and the subsequent effectiveness of detecting patients with low EF.

Prof K Srinath Reddy, who formerly headed the department of cardiology at All India Institute of Medical Sciences, New Delhi and is founder-president Public Health Foundation of India, said AI can help collate large data sets and distill them into diagnostic pattern recognition and management algorithms. It can thereby overcome gaps in prior clinical experience. This can be very helpful in promoting early recognition, prompt care and appropriate referral practices in primary care. “However, the algorithms developed by AI are dependent on the representativeness and accuracy of the input data. They are context-dependent and algorithms developed in a Western populations are not automatically always applicable in the Indian context. So, we need to develop our own AI capabilities, using large Indian data sets. It is clear, however, that those who can adapt well to use of AI can improve their diagnostic accuracy and provide better patient care. This is especially so in primary care settings where care providers do not have specialist training,” he said.

Dr Sanjeev Jadhav, Director of the Heart Lung Transplant programme at Apollo Hospital, Navi Mumbai, said that technology has helped in diagnosing and treating patients in various fields of medicine and surgery. “AI is still at a basic level and such tools generated in the context of the Mayo Clinic study will definitely evolve in the near future to help doctors diagnose right at the primary level whether the patients will have a heart problem or whether there is a problem with the heart function. Most patients approach the doctor when they have symptoms like heaviness in the chest or difficulty in breathing – AI will diagnose heart conditions before the symptoms start prompting early treatment. Technology will help doctors diagnose patients way before symptoms start and this will help in future,” Dr Jadhav said.

A total of 165 clinicians and 11,573 patients were included in the analysis. Clinical data was collected using electronic health records. A patient’s data was included in the analysis if the patient was 18 years of age or older and had received an ECG for any indication between August 5, 2019 and March 31, 2020. Only the first ECG of an individual patient was considered for the decision to order an echocardiogram within the study period. Patients’ data were excluded if they had known that the EF was less than or equal to 50 per cent or had a history of heart failure before the ECG. The study found that the proportion of AI-positive with confirmed low EF was 33 .9 per cent in high adopters and 16.3 per cent for low adopters.

Researchers have said that to impact human health, AI tools must be adopted by clinicians. Researchers in the study found a wide variation in the rate of adoption of AI recommendations. Those who responded to the AI and ordered an echocardiogram were significantly more likely to identify left ventricular dysfunction in their patients (33.9 per cent vs 16.3 per cent).

AI has promised to augment decision-making for clinicians in health care for decades, researchers said. Recent advances, adoption of electronic health records (EHRs) and the ability to apply machine learning to this enormous data repository have now made it possible to use AI to improve diagnostic accuracy and refine treatment plans, they added. They observed in the study that primary care clinicians, who were high adopters of an AI-enabled clinical decision support tool, were twice as likely to diagnose low EF than low adopters.
They also concluded that clinicians most likely to follow through with the recommendations of the AI decision aid tended to be less experienced in dealing with complex patients and this underscores the importance of clinician education and engagement and AI systems that integrate seamlessly into the workflows of busy caregivers.