BY JYOTHSNA HEGDE
Atlanta, GA, December 9, 2023: Since The Master Mystery in 1918 to The Creator in 2023, Robotics and Artificial Intelligence (AI), through the imagination of its creators have entertained and intrigued the masses. The reality of AI, however, is far more impressive and impactful. With high-level automation and intelligent machines coupled with powerful analytics and computer processing power, AI, even in healthcare, is reinventing and reinvigorating through machines that can predict, comprehend, learn and act, supplementing human intelligence to provide more sustainable and scalable solutions.
Unleashing the power it holds in the field of medicine, Emory University launched Georgia’s first institute of its kind – The Empathetic AI Health Institute this November with a key focus on the application of AI in the prevention, treatment, and diagnosis of diseases in Georgia and across the globe. Collaborating with multidisciplinary experts, the institute aims to leverage AI to save and improve lives, making care more proactive and personalized.
The Institute is part of AI. Humanity, a university-wide initiative dedicated to shaping the artificial intelligence revolution to better human health, generate economic value, and promote social justice. The three-pronged mission of the institute is to innovate, deploy, and scale up accessible, cost-effective, and equitable artificial intelligence (AI) tools into healthcare solutions for patients. To accomplish this, it seeks to establish a diverse network of experts from Emory, the Atlanta VA Medical Center, Georgia Tech, and beyond, while also forming public-private partnerships to accelerate research.
“Our focus and our mission within the institute is really on developing, and innovating novel advances in AI technology, specifically around health and medicine. We will be working with our partners at Emory healthcare. We’re then going to deploy this technology locally within the healthcare enterprise and set the stage for ultimately deploying and scaling this to technologies, to countries all over the world, including my own native country of India,” Dr. Anant Madabhushi remarked.
In his exclusive with NRI Pulse, Dr. Anant Madabhushi, the Executive Director for the Emory Empathetic AI for Health Institute shed light on vision of the institute and the immense potential of AI to discern patterns in vast amounts of data and make predictions that improve patient health outcomes in diseases such as lung, prostate and breast cancer, heart disease, diabetes and more, emphasizing health equity by making care more affordable and accessible. Dr. Madabhushi also highlighted the partnership with India under this umbrella.
Dr. Madabhushi holds a primary faculty appointment in the Department of Biomedical Engineering, and secondary appointments in the Departments of Radiology and Imaging Sciences, Biomedical Informatics (BMI) and Pathology.
The phrase “artificial intelligence” was first coined in a Dartmouth College conference proposal in 1955. But the AI applications did not enter the healthcare field until the early 1970s when research produced MYCIN, an AI program that helped identify blood infections treatments. The proliferation of AI research continued, and in 1979 the American Association for Artificial Intelligence was formed (currently the Association for the Advancement of Artificial Intelligence, AAAI).
The inspiration behind Dr. Madabhushi’s work in the field of AI and cancer is deeply personal. “I lost my aunt to breast cancer in India. One of the things that I’ve been very acutely tuned to is how can we leverage the power of technologies like this to have an impact, particularly in low middle income countries like India, where you don’t have access to more expensive, sophisticated technologies,” he expressed.
Vision of The Empathetic AI Health Institute
“The goal here is to leverage the power of artificial intelligence and machine learning and use it with routinely acquired data with radiology, pathology, and really bring precision medicine in an affordable equitable way to patients all over the globe. We want to develop tools that are going to be affordable, accessible and equitable. At the same time, we want to focus on innovation, deployment, scaling and translation,” Dr. Madabhushi said. “And at Emory we have an opportunity to really lead the charge in the space because as a unified system where we have a top-class university, along with an amazing healthcare enterprise, we have the opportunity not just to do the innovation, but then also to be translating the projects into clinical deployment.”
“Thanks to Provost Bellamkonda there has been a focus over the last few years to really transform Emory into the hub for AI and it’s under the broad vision a broad umbrella of AI humanity, where we are really seeing the opportunity for impact with AI across multiple different streams right across the social sciences, humanity, law and multiple disciplines,” he continues. “ The Emory empathetic AI for Health Institute falls under this umbrella of AI humanities, but with a focus on health with the focus on medicine. And the mantra, is that we will continue to push within the institute is around affordability, accessibility and equity so that we can deliver the promise of affordable, accessible equitable precision medicine to everyone not just in Georgia, but ultimately all over the world.”
The Promise of AI
“We’re not trying to replace the physician. We’re trying to provide them with the tools that will allow them to do their job even better,” Dr. Madabhishi said about AI. The promise of AI, he expounded, was its ability to mine through large amounts of data and find subtle cues and patterns that relate to disease diagnostics which allow for better and early detection, better characterization, and to do it in a way that is potentially more robust and resilient compared to human interpretation. The other promise of AI beyond diagnosis, he noted, is being able to identify the aggressiveness of the disease.
In the United States, approximately 40% of the adult population will be identified or diagnosed with some form of cancer in their lifetime. And not every cancer needs to be treated because a lot of them are actually fairly indolent, and not very aggressive. And the big promise for AI, Dr. Madabhushi observes, is the way “to use these tools with routine radiology, routine pathology, to allow for better characterization.” In other words, AI can identify those patients who could benefit from just a watchful waiting approach, with no intervention versus those patients who have far more aggressive disease, and therefore need intervention right away and therefore really move precision medicine forward,
“We know that a lot of our patients will develop diseases like cancer and other chronic conditions.” The power of AI, Dr Madabhushi notes, is its ability to uncover treatment strategy that is personalized to each patient.
Citing immunotherapy that has changed the landscape of treatment of cancer patients, particularly in countries such as India, Dr. Madabhushi said the cost – somewhere on the order of a quarter of a million dollars per patient per year, putting it well beyond the reach of most patients. And sometimes despite payment immunotherapy does not work on certain patients. AI in such cases can be effectively used to predict patient responses to these drugs, and therefore identify the appropriate treatment regimen for these patients.
Citing another example, Ductal carcinoma in situ (DCIS), a non-invasive or pre-invasive breast cancer that occurs when abnormal cells are found in a breast milk duct, Dr. Madabhushi noted that while surgery is the first intervention, the question of the requirement of radiation therapy post-surgery remains. “The truth is that only a subset of women will actually benefit but we don’t have a good way of figuring that up.” With AI, he said, biopsies of the DCIS can be analyzed to predict which women are going to receive the added benefit from radiation therapy versus those women who can avoid the radiation therapy and will benefit from surgery alone. And the same has been found to be true in the context of head and neck cancer too, where the majority of patients receive radiation therapy, but AI can identify those patients who could benefit from reduced radiation dose, and be perfectly fine.
The other piece of AI research in the context of health, Dr. Madabhushi notes, is the differences across populations. With his roots from India, he said his research is always geared towards a global outreach, ensuring that diagnosis, treatment regimens, and patient responses are accounted for everyone across the board. “I think deeply about whether the AI tools that we’re developing are really going to work across different populations. And what our algorithms are already revealing is that there are subtle differences in the appearance of the disease across diverse populations.” His research team in fact has put out some data a showing breast cancer is slightly different between South Asians, versus a not North American population of European ancestry or even African American ancestry. The promise of AI then is to account for these subtle differences and create more tailored, more precise, more accurate population specific models, he says.
Why not humans?
“We know that humans are great in rendering diagnosis both on radiology as well as on pathology, images. The problem is that humans tend to tire there is a lot of variability between interpretations across different humans. And so the ability to have AI algorithms that can read in a more consistent manner will allow for more reproducible robust decision support.”
“We also have to think about affordability. We have to think about accessibility, and we have to think about equity. We have to make sure these tools are really working across diverse patient populations.”
While possibilities are aplenty with AI, Dr. Madabhushi notes, the key is to identify the right opportunities, application areas, and the right cohorts.
Affordable, Accessible and Equitable Healthcare Solutions with AI in the US and across the Globe
“We don’t want to be in a situation where you can develop very sophisticated complex technology, but it’s only going to work for those who can afford it,” Dr. Madabhushi notes. “Ultimately, AI has to be developed with the context of different kinds of data, across diverse backgrounds, so that we can start to find patterns that tell us about risks, smart outcomes and treatment response.” So I think that’s really critical, really important consideration here.” And then, he continues, “we have to be working with implementation scientists – people who really understand implementation, who are able to really take the technology and the innovation that we’re doing and deploy it in a widespread way.”
“One of the things that we’ve been very focused on within our group is leveraging the power of AI with routine data,” Dr. Madabhushi said. Routine data, for cancer patients for example, includes the mandatory radiology scans, and pathology images. By leveraging the power of these radiology pathology images (and other routine data), AI can find patterns and features that can identify who is going to benefit from the more aggressive treatments verses those that don’t and predict outcomes for these patients. But because AI uses routine data, there is no additional cost.
Citing breast cancer treatment that killed his aunt as a example, Dr. Madabhushi said chemotherapy can be avoided in cases diagnosed early on. But the only way of really being able to identify that is with a molecular base test, a genomic based test one that involves destructive testing of the tumor tissue and looks at the expression of different genes. Apart from involving destructive testing of the tissue, the shipping of the tissue to a specialized lab in Redwood, California, where it’s performed costs $4,000.
“What we have shown very recently in a publication is that when you use AI with biopsy images, we can come up with predictions that are potentially even better than what you’re getting with this $4,000 test!” Dr. Madabhushi observed, adding, “essentially, you’re doing it for pennies on the dollar because we’re not acquiring anything outside the clinical workup. We’re just talking about an image of a biopsy every patient will have every cancer patient is going to have a biopsy.”
This context also plays a huge role in countries like India where cost of chemotherapy is about $1,900, he noted and adding that there is no justification when the cost of that drug is half the cost of the test that tells you whether or not the patient needs it.
“That’s where having tests that are really affordable particularly for low middle income countries and rural areas in the US becomes so important because we have to think about affordability, accessibility and equity.”
Collaborations and Partnerships with India
During Indian PM Narender Modi’s state visit to the US upon President Biden’s invitation, leaders welcomed the launch of a $2million grant program under the U.S.-India Science and Technology Endowment fund for the joint development and commercialization of Artificial Intelligence (AI) and quantum technologies and encouraged public-private collaborations to develop high performance computing (HPC) facilities in India.
The AI in healthcare market is expected to hit around USD 3,55,780 million by 2032, increasing from USD 15,100 million in 2022, and growing to CAGR of 37.66% from 2023 to 2032. AI in Indian Healthcare Market is predicted to grow from $14.6 billion in 2023 to $102.7 billion by 2028.
A report on the National Strategy for Artificial Intelligence by the NITI Aayog showed that shortage of qualified healthcare professionals and non-uniform accessibility to healthcare across India. India has only 64 doctors available per 1,00,000 people compared to the global average of 150.
Recognizing this need, and being a ‘mumbaiwalah,’ per his description, Dr. Madabhushi said it was important to him to create collaborations and partnerships with India to make an impact there. True to his mission, Dr. Madabhushi highlighted two important collaborations between Tata Memorial Cancer Center in Mumbai and L V Prasad Eye Institute in Hyderabad.
The collaborations between Tata Memorial Cancer Center to develop AI technology for low-cost diagnostics for breast oral and prostate cancers, he said yielded fruitful results with funding firstly through a US indoor joint grant and recently through National Cancer Institute in the area of oropharyngeal cancers or head and neck cancers.
At the L V Prasad Eye Institute, AI technology is being applied in the context of diabetes. “We are looking at diabetic conditions of the eye, diabetic retinopathy and diabetic macular edema, to be able to predict which patients are going to respond to certain treatments and which patients are not.”
Dr. Madabhushi said he was recently in Mumbai for the Indian cancer Congress to “not just discuss the research, but really thinking about that next step about implementation and deployment.”
The Challenges
While there are some business challenges such as getting reimbursement codes and regulatory approvals , Dr. Madabhishi says to him personally, the biggest challenge and responsibility of AI scientists is “being able to get people to use these technologies, getting physicians on board, and then being able to implement it and deploy it at scale, so that the maximum number of people can benefit and doing it in a way that benefits all populations. “
The other challenge, he notes, is the presentation of business case for companies developing these tools. While Implementation poses few challenges, he notes that interpretability is important too, because physicians must be convinced to have confidence in the tools and must be willing to use it.
The word “empathetic” in the Institute’s name is a reflection of how much the quality of empathy and the challenge of achieving it matter at Emory. The Institute’s leaders know that American health care outcomes still show big disparities between different groups and they’re focused on working to make AI models more inclusive. “As we think about AI,” Dr. Madabhushi concluded, “we need to make sure that we’re imbuing that same sense of empathy in the development and the application of AI tools for precision medicine.”
—
About Dr. Madabushi
Prior to joining Emory, Dr. Madabhushi was at Case Western Reserve University, where he was the director of the Center for Computational Imaging and Personalized Diagnostics and Donnell Institute Professor, Department of Biomedical Engineering. He was also a research health scientist at the Louis Stokes Cleveland Veterans Administration Medical Center. He has authored more than 450 peer-reviewed publications and holds more than 100 patents either issued or pending in the areas of artificial intelligence, radiomics, computational pathology, medical image analysis and computer vision. His work on “Smart imaging computers for identifying lung cancer patients who need chemotherapy” was called out by Prevention Magazine as one of the top 10 medical breakthroughs of 2018. In 2019, 2020 and 2021, Madabhushi was named to The Pathologist’s Power List of 100 inspirational and influential professionals in pathology. Dr. Madabhushi has secured almost $80 million in grant funding and co-founded three companies. In addition, more than 30 technologies developed by Dr. Madabhushi’s team have been licensed.