Using AI can improve health outcomes for women
IMPACT Lab develops tools for women’s heart health and maternal care
Cardiovascular disease affects more women than all forms of cancer combined. It is the leading cause of death, killing one in three women annually. It is also responsible for more than 30% of maternal deaths.
But when women seek treatment for heart conditions, they often receive care based on studies conducted on men.
“The need to improve cardiovascular care outcomes for women is probably one of the most pressing health issues worldwide — and certainly in the U.S. and India,” Sandeep Gupta says.
Gupta is a professor of computer science and engineering in the School of Computing and Augmented Intelligence, part of the Ira A. Fulton Schools of Engineering at Arizona State University. He also heads the Intelligent Mobile & Pervasive Applications & Communication Technologies Lab, known as the IMPACT Lab, where he is working with Ayan Banerjee, a Fulton Schools research associate professor.
The pair is exploring applications of artificial intelligence, or AI, in the medical sector. They have received multiple grants, including from the National Science Foundation, to develop tech tools to improve treatment options for women.
Gupta explains that for decades the medical sector believed men were predominantly at risk for heart attacks and coronary conditions. Doctors — and female patients themselves — lack essential information about preventive care for women. Life-saving medications, such as statins, were developed using clinical trials that largely recruited pools of male subjects. Women more frequently suffer from adverse reactions to these drugs.
This approach created a world where women are less likely to suffer from heart disease but are far more likely to die from it.
“But AI holds great promise to revolutionize women’s health care by moving from a one-size-fits-all approach to personalized and precision medicine,” Gupta says.
A team making an IMPACT
They say everyone has a twin somewhere out there, but if Gupta and his team get their way everyone will also get a digital twin. The researchers are developing tools that use a form of AI known as machine learning. In these AI-based systems, software programs comb through, analyze and compare enormous sets of data, detecting patterns and drawing useful conclusions. In the School of Computing and Augmented Intelligence, machine learning systems are being used to virtually model new materials, discover new scientific laws and address social isolation in aging veterans.
The research team in the IMPACT Lab is creating new technology that will combine expert medical knowledge, available research data and information from a patient’s own health history to create a profile that can be used by doctors and health care professionals to provide individualized care. The goal is to harness the power of AI to fill in the gaps in current medical research.
He says that the cooperation his team is receiving from doctors and medical researchers at top U.S. institutes, including Phoenix Children’s Hospital and the University of North Carolina at Chapel Hill, is key.
“Solving these critical problems requires close collaboration between engineers, scientists and clinicians to come up with solutions that work,” Gupta says. “In the IMPACT Lab, we are taking this approach, using AI to ensure serious health issues for different population segments, including women, get adequately addressed.”
Because effective screening is one of the most important aspects of getting women at risk for coronary conditions the treatment they need, one of the first applications of IMPACT’s research is a reader for electrocardiograms, or ECGs. An ECG is a fast and simple test in which doctors assess the electrical activity of a heart.
The AI-enabled reader will automatically analyze the results of an ECG and provide the statistical likelihood that the female patient has coronary artery disease, or CAD. The goal of the project is to remediate gender disparity in CAD diagnoses. Gupta believes that the work will accomplish two important objectives. It will eliminate the need for unnecessary and stressful cardiac imaging in women at low risk and will also ensure that doctors have accurate information when making decisions about their care recommendations.
The IMPACT Lab researchers have conducted successful demonstrations of the system for teams of clinicians and are busy applying expert feedback.
Pregnant women with Type 1 diabetes face increased risks
The IMPACT Lab team is also tackling key issues in endocrinology.
Expectant mothers experience a whirlwind of emotions, from joy to nervousness to bliss to concern. Women with Type 1 diabetes have extra reasons to worry.
Type 1 diabetes is an autoimmune disease that often begins in childhood. The condition causes the pancreas to make little or no insulin, a hormone the body needs to convert sugar into energy. To prevent sugar from building up in the bloodstream, people with diabetes require an external source of insulin, often delivered in the form of a shot.
Normal hormonal changes expectant mothers experience during pregnancy can cause serious problems for women with diabetes. Blood sugar can fluctuate more widely for pregnant women with diabetes and excessive blood sugar is linked to adverse outcomes such as large birth weight babies who must be delivered via more risky Cesarean sections.
An automated insulin pump is a small wearable device often placed on the back of the arm that automatically delivers the correct amount of insulin at needed intervals. It is one good solution to the challenge of ensuring expectant mothers with diabetes get the insulin they need.
“But there are currently minimal industry-driven efforts to develop a commercial grade automated insulin delivery system for women with Type 1 diabetes who are pregnant,” Gupta says.
Gupta notes that the medical sector is understandably reluctant to conduct studies on pregnant women. The stakes there are high, and no one wants to be responsible for unfortunate consequences to babies or their mothers.
Again, he hopes AI can help.
The IMPACT Lab team is using AI to run simulations on how much insulin pregnant women with diabetes need and create algorithms that can be used by the pump to deliver the correct dose as needed. They are working with industry partners and hospital clinicians to create a pump prototype. The team can also use AI to run virtual tests, eliminating the need to involve pregnant women in potentially dangerous studies.
Moving forward, Gupta is excited to see this technology deployed in real-world settings. His team hopes to submit their automated insulin pump for approval from the U.S. Food and Drug Administration. They are also exploring applications of their cardiology projects with leading health organizations.
“Our goal is to create health care smart systems that can benefit understudied populations,” Gupta says.