How Artificial Intelligence Is Changing Cardiovascular Medicine
Adham El Sherbini, Investigative Journalist
December 4, 2022
Fear not — artificial intelligence (AI) isn't here to unleash widespread unemployment on the human population. Actually, to make up for the areas in which AI lacks, more humans will be needed.
According to Forbes, AI will, above all, assist with human jobs. Almost all professions, and especially those in the medical field, require an essential degree of human judgment.
Among AI applications in self-driving cars, retail for forecasting, and analyzing video surveillance, medicine is comparatively within its infancy stage. Thus far, however, AI has proven particularly effective in cardiovascular medicine.
Although AI's potential in cardiovascular medicine is promising, its ethical and legal limitations are exigent.
"[AI can] develop an algorithm capable of replicating a sector of human intelligence," said Dr. Alexander Wong, Canada Research Chair for Artificial Intelligence and Medical Imaging at Waterloo University.
Of its wide-ranging applications, machine learning (ML) is a sector of AI focusing on learning from data and using data to identify relevant patterns. ML has been particularly effective in making accurate predictions of diseases and outcomes.
AI is used for the mass detection of several cardiovascular diseases, including atrial fibrillation, heart failure, and cardiac arrest. Early detection allows for early intervention, ultimately leading to better outcomes.
Wearable technologies, such as Apple watches, have begun integrating AI to provide real-time updates based on cardiac activity collected by a watch or accelerometer.
Several studies have used cardiac health data from Apple watches to make predictions of heart diseases — such as in the early detection of atrial fibrillation.
"AI has been successful in the monitoring, prediction, and screening for cardiovascular diseases," said Wong.
Wong emphasized the benefits of AI for preventive cardiology – a field of medicine working on preventing cardiovascular diseases by monitoring risk factors such as drug use, smoking, blood pressure, nutrition, physical activity, and weight change.
However, practitioners rarely use AI due to legal and trust complications. The hypersensitivity of patient health records poses various legal issues for AI in clinical settings.
One of the main limitations following an AI diagnosis is that no sole stakeholder is responsible for the decision – previously, physicians held the legal responsibility.
Regarding standardization, there are no judicial methodologies to govern AI applications in medicine, permitting breaches and gaps in its implementation. Given the absence of legal responsibility and judicial rulings with AI, its integration into clinical practice would result in an influx of ambiguous court cases.
The legal ambiguity surrounding AI in medicine may lead to new jobs for lawyers and judges.
As AI lacks emotional intelligence, patients may not trust its diagnosis. Crucial components of the doctor-patient relationship are empathy and trust. Although most AI models have high accuracy rates, they do not explain the thought process behind a decision. Patients are more likely to question an algorithm's diagnosis — whose method is relatively opaque — than a human doctor.
AI poses many security issues due to hacking. Combined with the sensitivity of electronic health records, hacking could be catastrophic. Hackers may have a greater incentive, such as leaking health records. Integrating AI into cardiovascular monitoring technology involves third-party corporations — increasing the risk of security complications, as seen in the Cambridge Analytica scandal or Zoom hack.
Tech start-ups and large technological corporations (Google, IBM) are addressing AI concerns in medicine to ensure a safer and universal application. One way to mitigate the lack of trust is through developing a human interface to combat trust issues and would allow for direct interaction with patients in-person or through chatbots.
Although AI is making strides to advance the diagnosis and monitoring of cardiovascular diseases, it has yet to have the potential to replace physicians completely. With numerous concerns about trust, privacy, and ethics, humans will always be needed in medicine.