Artificial intelligence (AI) is here to stay. Like any tool developed by humankind, it can be used for good or to exploit. When applying AI to public health programs and response, great potentials exist for swifter, nimbler care and pin-pointed personalization. Its use also could exacerbate biases and the misuse of private information.
Dr. Hassan Tetteh, MD“Using AI safely, effectively, equitably, and ethically, it's only going to enhance the health of the population, the individual, and ultimately prevent disease in a way, on a scale that we haven't been able to do before,” says Dr. Hassan Tetteh, chief veterans’ health officer,
AIMindsystems Foundation and author of
Smarter Healthcare with AI: Harnessing Military Medicine to Revolutionize Healthcare for Everyone, Everywhere. “The main tenets of public health are protecting the public from harm and promoting health and wellbeing. That task alone is daunting for any one entity, any one office, or any one public health officer to administer on their own.”
Tetteh sees AI as a force multiplier for public health agencies as they work to promote health, prevent disease, and prolong life. On the individual level, public health can promote the use of devices, like smart watches that can collect data on heart rate, diet, activity, and sleep quality. The app on the device can then prescribe specific behavioral changes that improve the individual’s health — and, with consent, share that information with the person’s health care provider.
On the macro level, Tetteh sees AI as improving public health’s ability to predict and track disease, determine where exposure risks lie, and ensure treatments and vaccines are available.
“It certainly gives public health officials a lot more tools to be more effective in their jobs,” he says. “All of this goes hand in hand with the science and the art of preventing disease, prolonging life, and promoting health, which really is what public health is all about.”
When conducting research on the adoption of technologies in healthcare, Dr. Mary Brown, EdD, Steelcase culture and diversity consultant, concluded that predictive health, predictive analytics, genomics, and telehealth would fit well with the application of AI. Prior to working at Steelcase, Brown was learning and development consultant diversity, equity, and inclusion at Spectrum Health (now
Corewell Health).
Dr. Mary Brown, EdD“I saw the benefit of using AI in these particular areas — the disease tracking, the predictive analytics to look at health trends that are happening, organizing and allocating resources, and personalized interventions,” Brown says. “But we definitely need policies and safeguards that would support the positive side of the impact, as well as look at mitigating bias and inequities.”
These AI policies would ensure equity, mitigate bias, stipulate inclusive data collection, maintain transparency, and require regular audits. In public health applications, data used to fuel AI models must accurately reflect the populations being served. Policy must mandate the inclusion of diverse groups and demographics and incorporate protections for people who are often marginalized within Michigan’s communities. In addition, gathering data that does not reinforce stereotypes will require transparency on how AI algorithms are designed.
“A lot of the challenges that people have with AI are because they may not know how it works,” Brown says. “So, the AI model should be transparent and clear and should have documentation on how decisions are made. When you do that, that helps build trust.”
Dr. Sharon Kardia
Preparing public health students for an AI future
Recognizing the potential and problems that AI can usher into public health, the
University of Michigan (U-M) School of Public Health acknowledges the anticipated integration of AI modalities into public health with two masters-level courses:
Artificial Intelligence For Health Care Policy and
Chatgpt/AI And Public Health.
“This artificial intelligence emergence is here to stay,” says Sharon Kardia, PhD, U-M School of Public Health associate dean for education and Millicent W. Higgins collegiate professor of epidemiology. “It's better for us to move slowly knowing that it is going to be something that we have to take a step out into the unknown, gather some information about how well it works, and then keep expanding.”
Kardia teaches the
Chatgpt/AI And Public Health class. Course learning objectives include articulating human values driving public health and describing how AI tools could undermine them; identifying unique challenges and opportunities AI poses; and using critical AI literacy to develop best practices for incorporating AI into public health work.
“There are many different reasons why artificial intelligence is important to public health students. One is that the technology itself has been known for quite a while,” Kardia says. “AI could help us do a better job of figuring out how we might get in front of the next pandemic, things that are as big or as societally important as that.”
Early in the COVID-19 pandemic before vaccines were available, Tetteh and his colleagues at the
Joint Artificial Intelligence Center, a subdivision of the Department of Defense, embarked on
Project Salus.
“We were like many other people, feeling handicapped and helpless, because we were all at home dealing with our own fears and anxiety, yet we felt this urgent need to do something,” Tetteh says. “It was many, many late nights, conference calls, a lot of brainstorming, and virtual whiteboarding that led to a connection of ideas and basically a theory that we could use data that was definitely in the domain of availability and engineer a model.”
That
model used AI to analyze 11 million pages of Medicare claims data to predict areas of the country that would see the most COVID-19 cases with high accuracy.
Detroit was one location identified. Hence, Michigan saw an influx of National Guard troops helping out as well as increased access to masks, ventilators, and, when they became available, COVID vaccines.
“There are elements of that model and the platform that are being used in different capacities for different use cases and for different applications, but not in the way it was used at the time of the pandemic,” Tetteh says.
The U-M School of Public Health’s course will help future public health professionals understand the important benefits that AI can bring to the field — as Project Salus did during COVID — as well as its potential risks. Kardia explains that using AI to develop biostatistics, create predictive modeling, and expedite literature reviews could make dealing with the next pandemic far less formidable. She sees AI taking on the role of personal assistant as another benefit for public health, which, like most health care professions, is experiencing a serious workforce shortage. AI can also enhance public health responses by summarizing information quickly, speeding up data analysis, and improving communication.
“How do you use AI to do something like that, not in six months, but in six days? That pulls us into a different kind of world,” Kardia says.
U-M School of Public Health instructional design consultant Frederique Laubepin and PH555 student Mahalakshmi Sivashankaran. Laubepin helped design the PH555 course.
Ethical, responsible use of AI
All master’s level U-M School of Public Health students sign a pledge to use AI responsibly. That includes considering its environmental impact on water and electricity as well as upholding professional ethics, for example, citing the use of AI summaries.
“Using citation of AI is a way for there to be transparency, which is another key ethic in public health. People can see what was done, how it was done, and either replicate it or refute it,” Kardia says. “The pledge is meant to make people more mindful of what they're doing and how they're doing it.”
Kardia, Brown, and Tetteh agree that policies and regulations must ensure AI's accuracy, reliability for all demographics, and human oversight.
“We have a whole range of concerns and hopes that go with some of these technologies. We're going to have to think in new ways on how to evaluate it,” Kardia says. “Let's say, somebody builds a mental health bot and, inside of it, are ads aligned with pharmaceutical companies … the detection of bias or the detection of potential harm is going to be tricky.”
However, AI can enhance public health responses by personalizing care, summarizing information quickly, aiding in data analysis, developing vaccines more quickly, and improving communication.
“From a very holistic standpoint, what AI can do is help to facilitate the work that public health officials and their agencies and departments are charged with,” Tetteh concludes. “AI is going to continue to make public health officials more effective in their work and, if used effectively, empower individuals to take more agency over their own health and wellbeing.”
Estelle Slootmaker spends most workdays as a journalist and book editor. She also writes poetry and will be publishing a new children’s book, “Places Where the Sun Don’t Shine.” You can contact her at [email protected] or www.constellations.biz.
Photos by Doug Coombe.
Photos of Dr. Hassan Tetteh and Dr. Mary Brown courtesy subjects.
The Yours, Mine, and Ours — Public Health series highlights how our state's public health agencies keep us healthy, safe, and informed about issues impacting physical and mental health in our communities, homes, workplaces, and schools. The series is made possible with funding from the Michigan Association for Local Public Health.