Artificial intelligence could be key to detecting diabetes and heart disease among South Asians

Ravi Parikh MD MPH, Chairman Manhattan Retina and Eye Faculty, New York University School of Medicine. Photo: Anne Rayner, Vanderbilt

According to a study by the NYU Center for the Study of Asian American Health (CSAAH), diabetes, which increases the risk of heart diseases, is rife among South Asians. The report’s findings conclude that in the United States, South Asian immigrants are 7 times more likely to have type 2 diabetes than the general population, and in New York City, Indian immigrants are at a greater risk of hospitalization for diabetes than other immigrants.

The CSAAH study found that approximately two-thirds (67%) of South Asian CHNRA (Community Health Resources and Needs Assessment) respondents had ever been screened for diabetes, and 17% of those screened had been told by health professional that they had diabetes. This rate of diabetes is almost 3 times higher than the rate for Asian Americans living in NYC (6%).
The CSAAH study also noted a 2005 analysis of 1.5 million New York City birth records registered between 1990 and 2001, which showed South Asian women experienced the highest prevalence rate (11.1%) and the highest increase for gestational diabetes (95% increase since 1990) when compared to other groups.
A March 31, 2020 study by K.M. Venkat Narayan & Alka M. Kanaya, found that South Asians have prevalence of type 2 diabetes even at a lower body-mass-index (BMI), which may have to do with a lower ability to secret insulin.
The University of California, San Francisco did a study released in 2016 which examined South Asian heart disease risks, entitled “Mediators of Atherosclerosis in South Asians Living in America” or the acronym pun MASALA, the first study where researchers collaborated with colleagues in India.
It revealed that subjects in both India and in the U.S. are at risk of cardiovascular disease, but that those who became more bicultural had a lesser thickening of arteries. It also found that current guidelines set by the American Diabetes Association for screening for type 2 diabetes and measuring BMI, could not be applied to South Asians who needed a lower threshold. So these researchers recommended the ADA change the screening guidelines for Asian Americans.
With diabetes and heart disease being so common among South Asians, researchers have said artificial intelligence (AI) could play a big role in diagnostics for these specific diseases. AI devices could assist in early identification or diagnosis of a condition without a physician, and prompt higher risk or patients earlier in their disease course to seek care before catastrophic complications.
The first autonomous artificial intelligence diagnostic device approved by the Food and Drug Administration (FDA),IDx-DR, is able to detect diabetic retinopathy (the leading cause of blindness among working age adults in the US) by capturing an image of the eye and having its AI software analyze the image for signs of disease without a physician.
Dr. Ravi Parikh MD MPH, Chairman of Manhattan Retina and Eye Consultants and Faculty at New York University School of Medicine has published about this technology and noted its potential and notes, “many individuals living with diabetes see their general physician but may not see an ophthalmologist or retina specialist to evaluate their eyes. Having AI technologies at their general physician’s office or pharmacy may be a way to evaluate and identify more patients and bring those who need sub-speciality care in for treatment before irreversible and serious complications such as blindness occur.”
Similarly, AI technologies for heart disease and other illnesses could potentially be used to identify at risk patients earlier and bring them into subspecialty care sooner preventing more serious disease. For example, he points to already existing AI technology to enhance diagnostic quality in cardiac care units, such as EchoGo and EchoGo Pro, award-winning AI technologies that automate cardiac ultrasound measurements for heart functions and make it possible for physicians to even predict occurrence of coronary artery disease.
Even though the Food and Drug Administration (FDA) has recognized the importance of AI through its expedited approval plan of IDX-DR, there are many ecomonic challenges to the widespread adoption and utilization of AI, hindering AI’s potential to improve patient care.   .
In their article, entitled, “Economic Challenges of Artificial Intelligence Adoption for Diabetic Retinopathy,” carried in Ophthalmology, journal of the American Academy of Ophthalmology,
Dr. Parikh and co-authors argue for making AI economically viable through numerous steps – developing AI-specific reimbursements ; uniform insurance coverage;, and work flow considerations when implementing a new technology. Challenges may occur to universalizing the use of this AI device (IDx-DR) for diagnosing retinopathy because of the economic hardships clinics and hospitals would face if insurers and Medicare do not properly reimburse the use of such technology, Parikh told News India Times.
In Parikh’s paper, he and co-authors noted that despite the American Academy of Ophthalmology recommending  annual eye examinations for everyone with diabetes, some studies estimate only 15.3 percent of Americans with diabetes meet recommended guidelines despite evidence that examinations prevent vision loss and the financial hardships associated with vision loss from diabetic retinopathy.
“The government and insurers have to properly incentivize the use of these new technologies which have the potential to address health disparities by identifying patients  and having them treated earlier,” Dr. Parikh said.
Dr. Parikh noted that AI devices like IDx-DR would be even more important for South Asians keeping in mind that diabetes, like heart disease, occurs at disproportionately higher rates among South Asians.
The company that sells the device, charges per patient evaluated. If the cost to a clinic or
hospital is $25 per patient evaluated, as it is currently, and the reimbursement on average is less than that, or if the patient has diabetes but is negative for retinopathy and an insurer denies coverage for the evaluation, a clinic or hospital has no incentive to invest in such technology. Further, placing additional costs on patients may also limit AI use while also creating potential financial hardships on patients  Dr. Parikh contends.
Furthermore, “Apart from the cost of the device, it could also disrupt a general physician’s work flow in the office and affect the patient experience with potentially additional wait times and office visits from positive test results,” Dr. Parikh said.
“But everybody with diabetes needs a retinal exam,” emphasizes Dr. Parikh, and there lies the dilemma, which occurs in many other areas of medicine as well.
“AI will change medicine and medical innovations using AI are one way we can address health disparities. However, for AI to reach its potential in improving care and reducing disparities in heart disease, diabetes, or other conditions, physicians, clinics, and patients must be supported. Medicare and the government will have to incentivize AI use for its widespread adoption and for the benefits of AI to reach their full potential in helping patients,” Dr. Parikh said.


Please enter your comment!
Please enter your name here