Chasing the holy grail: How AI can transform diagnosis and intervention in healthcare

Empowering doctors with AI: Technology can redefine and transform healthcare through predictive analytics and diagnosis.

BySumit Jha

Published Jun 22, 2023 | 9:00 AM Updated Jun 22, 2023 | 11:51 AM

Chasing the holy grail: How AI can transform diagnosis and intervention in healthcare

It looks like a scene straight out of science fiction — or one from the James Bond franchise — where the audience expects the unexpected.

Consider this scenario: You visit an ophthalmologist for an eye scan. A scan using Artificial Intelligence (AI) later, the healthcare professional refers you to a cardiologist.

The AI scan has detected you with a cardiovascular condition, and considering the urgency of the situation, emphasised the need for immediate medical attention.

Impossible, do you feel? Hardly, it seems with scientists in hot pursuit of this possibility.

AI technology has the potential to revolutionise the healthcare industry. Its ability to mine and process vast amounts of data, predict outcomes, and assist in diagnosis and treatment decisions can significantly benefit doctors and patients.

Numerous doctors are open to embrace technology that can assist them in their profession.

Pichai’s tech pitch

During the Google I/O — the annual developer conference — in May, Alphabet Inc. CEO Sundar Pichai, while highlighting the impact of AI on healthcare, specifically dwelt on the use of deep learning in diagnosing diabetic retinopathy.

Sundar Pichai said AI systems have provided additional insights by detecting details that humans might have overlooked during analysis of retinal scans. Pictured, Pichai at the Google I/O meet on 11 May. (blog.google)

Pichai also emphasised that AI systems have provided additional insights by detecting details that humans might have overlooked during analysis of retinal scans.

“Your eye scan turns out to hold information with which we can predict the five-year risk of you having an adverse cardiovascular event, heart attack, or stroke,” the Madurai, Tamil Nadu-born technocrat said.

“So to me, the interesting thing is that you know, more than what doctors could find in these eye scans. Machine Learning (ML) systems offer newer insights,” he explained.

He further added that this could be the basis for a new non-invasive way to detect cardiovascular risk.

Doctors often face challenging situations. Getting advanced notice, say, 24 to 48 hours before the deterioration of a patient’s health condition, can greatly impact the outcome.

“Leveraging Machine Learning systems, we have worked with our partners to analyse vast amounts of medical records — over 100,000 data points per patient — exceeding what any single doctor could manage,” Pichai elaborated.

“As a result, we can quantitatively predict the likelihood of readmission earlier than traditional methods, granting doctors more time to intervene,” he added.

Also read: Doctors react to cases of sudden cardiac arrest in teens

AI and healthcare 

“I was unaware that diabetes could cause blindness. I used to ride a two-wheeler until I had blurred vision in my left eye. After eight months, I lost sight in that eye,” Elumalai, a patient at Sankara Nethralaya in Chennai, testified in a Google study.

Four years ago, researchers from Google, Aravind Eye Hospital, and Sankara Nethralaya in Chennai collaborated for a mission.

The mission was to develop an automated tool capable of detecting diabetic retinopathy, a prominent cause of blindness. Their groundbreaking algorithm was designed to swiftly analyse retinal photos and provide a diagnosis within seconds.

Soon, this algorithm is expected to gain autonomy, revolutionising the field of eye-disease detection and management. However, AI’s capabilities extend even further.

Earlier this year, Google introduced an algorithm with the ability to identify a person’s gender, smoking habits, and predict their five-year risk of a heart attack, solely based on retinal imagery.

Even other researchers have discovered that AI-enabled imaging of the veins and arteries in the retina can accurately assess the risk of cardiovascular disease, cardiovascular death, and stroke.

This breakthrough has the potential to provide a highly effective and non-invasive testing method for individuals at medium- to high-risk of heart disease, eliminating the need for clinic visits.

The findings, published in the British Journal of Ophthalmology, demonstrate that the AI tool effectively utilised data such as smoking history, hypertension medication, and previous heart attacks from study participants.

The researchers highlight that AI-enabled vasculometry risk prediction is automated, cost-effective, and non-invasive.

It could potentially reach a larger section of the population, thanks to its availability in community settings, such as high street locations, and its independence from blood sampling or blood pressure measurement requirements.

Also read: Study links keto diet to risk of heart disease; doctors agree

Doctors sceptical

Speaking to South First, doctors suggested that integrating AI with healthcare infrastructure would be beneficial. However, it is important to note that AI will not replace any diagnostic tools: Instead, it will enhance disease-prediction capabilities.

For instance, referring to the Google I/O event where Pichai spoke, author Harinder S Sikka tweeted that an eye scan can predict cardiovascular events.

“Good bye to CT Scan, MRI, Xray. Cardiovascular events can be predicted by eye scan. Doctors can now get clear view of what is inside the body of a patient. Sundar Pichai, Google AI [sic],” Sikka tweeted.

However, doctors were sceptical. “MRI and chest X-ray were never used to predict cardiac events. CT calcium scoring is the one among the above three, which was used to predict coronary risk,” Dr Mukharjee Madivada, Senior Interventional Cardiologist and Managing Director of the Pulse Heart Center, said.

“Risk prediction works like this: The algorithm will predict that you will have, say, 7.5 percent risk of cardiac event over the next 10 years. This may not be accurate,” he added.

Dr Sudhir Kumar, a Neurologist at Apollo Hospitals, Hyderabad, referred to a popular saying. “In the present era and in future, doctors who acquire knowledge and skills in AI will have a significant advantage and thrive in their profession,” he told South First.

“On the other hand, doctors who do not embrace AI or fail to utilise its capabilities may face obsolescence or struggle to perform optimally,” he opined.

“Though this technology is still in its infancy, it has the potential to identify signs of diseases like diabetes, hypertension, and high cholesterol at the earliest stages, allowing for timely intervention and improved outcomes,” he pointed out.

“Moreover, if the technology can be made more accessible and affordable, it could alleviate some of the workload of doctors and be used by other healthcare professionals, such as nurses, effectively optimising the doctor-patient ratio,” Dr Kumar added.

Also read: 85.7 percent of Keralites are not aware of stroke symptoms, finds study

Need for tests

Dr Madivada said that even though Pichai claimed that that AI is demonstrating a better model to predict events, its accuracy has to be tested. “It takes time, because events are usually measured over a 10-year period.”

“Even if the technology is ready and is better equipped to predict events than existing models, what it will (complement rather than) replace are the history, weight and height measurements, lipid profile, BP and sugar tests and noninvasive testing,” he said.

He added that another problem with risk prediction models is with the information. “Let us say that it predicts a patient is having 12.225 percent risk of a cardiac event over the next 10 years. What to do with the info,” he wondered.

“You need to do what has always been told. Risk reduction by modifying risk factors, healthy eating, regular exercise, and reducing stress. AI will definitely help in diagnosing and treating heart diseases. But the day is not here yet and progress has been slow and deliberate,” Dr Madivada said.

He added that a retinal scan is nowhere near being a holy grail.

“It will not tell you if you have a heart attack. It will not show the blocks in the blood vessels. It will not tell you the heart function. It will not tell you how your valves are. It will not tell you if you need medication, angioplasty or bypass surgery,” he explained.

“It will ‘predict’ the ‘possibility’ of you having a heart disease. You can then get conventional testing done to confirm or rule out that possibility,” he said.

Senior Interventional Cardiologist Dr Deepak Krishnamurthy said that prediction of disease risk is not the same as replacing diagnostic modalities.

“People are almost always confused between ‘risk prediction’ and ‘diagnosis of existing disease’,” he said.

Also read: Hypertension strikes 1 in 3 adults in Kochi’s urban slums

Boon to healthcare

Dr Sudhir Kumar said that AI technology has numerous benefits in the field of healthcare. It will provide a positive impact as it allows doctors to access vast databases of medical literature quickly, enabling them to diagnose and treat rare diseases more effectively.

“While human memory, including that of doctors, is limited, AI can process and analyse large amounts of information within seconds or minutes, saving valuable time. Therefore, doctors must embrace AI to stay updated and provide the best possible care for their patients,” Dr Kumar said.

AI can assist doctors in various ways. For instance, it can help identify drug interactions, especially for less commonly used medications, and provide insights into new treatments and precautions.

It can also aid in determining appropriate dosages for specific patient conditions, such as paediatric, pregnancy, liver, or renal diseases. Additionally, AI can help doctors identify common and uncommon side effects of medications by accessing the latest literature, allowing them to address patient concerns accurately.

Furthermore, AI can contribute to improving patient care by predicting readmissions and identifying individuals at higher risk of developing certain diseases.

“By analysing multiple parameters, such as age, gender, medical history, and various test results, AI can assist doctors in making more accurate predictions and providing proactive care,” Dr Kumar said.

“This predictive ability can significantly benefit doctors in managing patients and taking preventive measures to avoid complications,” he added.