A New Prescription for Pain: AI and Mindfulness

PREVIOUS STORY
Worcester Polytechnic Institute to Celebrate the Inauguration of President Grace J. Wang, PhD

NEXT STORY
Making History: WPI Brings Home Two NCAA Division III National Championships in Two Days

Worcester Polytechnic Institute (WPI) will lead a five-year study to determine whether artificial intelligence can help doctors steer people dealing with chronic pain away from potentially addictive opioids and toward mindfulness-based approaches.  

 The new National Institutes of Health (NIH) HEAL (Helping to End Addiction Long-term) initiative-funded study will employ machine learning, a form of artificial intelligence, to look for clues in patient data that could help doctors better determine who is likely to benefit the most from mindfulness-based stress reduction, or MBSR, in managing their pain. 

“For physicians, it will be a new day,” said Jean King, the Peterson Family Dean of Arts and Sciences at WPI. “To be able to predict who would respond well to non-pharmacological interventions will truly save lives.”  

WPI has received $1.6 million in NIH funding to start designing the trial; if the team’s defined benchmarks are met, the research team and the university could receive a total of nearly $9 million in research funding over the course of the next five years. 

The findings of the study could give healthcare providers powerful tools to help people avoid taking opioids that can lead to lifelong struggles with addiction. Over-reliance on opioids for pain management can have devastating consequences; in 2021, more than 16,000 people died from prescription-opioid-related overdoses, and more than 80,000 people died from overall opioid-related overdoses, one death every 6 minutes. There have been concerning increases in opioid-related deaths in Black and Native American populations. 

At the same time, chronic pain is also a major concern. A recent U.S. Centers for Disease Control and Prevention Morbidity and Mortality Report estimated that more than 51 million people–more than 20% of U.S. adults–have chronic pain.

Researchers

Previous studies have found that MBSR is effective in helping people deal with chronic pain, but the mindfulness-based approach does not work for everyone, and doctors and clinicians don’t know exactly for whom it would work and why.

Focusing specifically on chronic lower back pain in diverse populations, the study will glean physiological data such as sleep patterns, heart rate, and general physical activity collected via fitness sensors worn by 350 participants during a six-month trial. Combined with self-reported information on depression, anxiety, pain, and levels of social support, the data will be analyzed by custom-designed machine learning models to detect patterns that might be impossible for a doctor to notice. The information will allow the model to predict whether a patient would beneficially respond to mindfulness, helping doctors better tailor treatments for individual patients. 

That predictive power could prove to be a powerful tool for physicians who previously may have been wary of prescribing mindfulness-based stress reduction, said Carolina Ruiz, the WPI Associate Dean of Arts and Sciences and Harold L. Jurist ’61 and Heather E. Jurist Dean’s Professor of Computer Science, who has been involved in researching and teaching machine learning for more than two decades. She added that the machine learning model used in the study will be interpretable–doctors and researchers will be able to pinpoint exactly why a patient may or may not respond well to mindfulness methods.   

“It will save time for the patient—they won’t have to go through a treatment that isn’t going to help,” she said. “It will also save a lot in healthcare costs and could be applicable to other types of pain and other types of treatment.”  

The study, dubbed Integrative Mindfulness-based Predictive Approach for Chronic low back pain Treatment, or IMPACT, will bring together a diverse group of researchers at WPI, UMass Chan Medical School, and Boston University Chobanian & Avedisian School of Medicine. Along with King and Ruiz, WPI faculty researchers include Emmanuel Agu, the Harold L. Jurist ’61 and Heather E. Jurist Dean’s Professor of Computer Science and MPI, Angela Incollingo Rodriguez, assistant professor of psychological and cognitive sciences, Zheyang Wu, professor, mathematical sciences, and Benjamin Nephew, assistant research professor, biology and biotechnology.  

Meet the team

Ben Nephew 

Ben Nephew

Assistant Research Professor, Biology & Biotechnology 

A behavioral neuroscientist, Nephew works closely with Jean King, studying the neurobehavioral mechanisms of mental illness and developing effective interventions.  

Nephew will oversee research workflow and administrative tasks and will serve as a “conduit between the clinicians and the machine learning folks on the grant.” 

The potential for the grant to identify potentially life-saving alternatives to managing pain is exciting, Nephew said. Creating targeted therapies has been highly successful in treating childhood cancer; adapting that approach to chronic pain is a relatively new idea, he noted.  

“The same efforts haven’t been applied toward behavioral therapies, or even mental illness, even though mental illness is highly correlated with chronic pain—what it does to your mental health, and how functional you are.”  

When he first partnered with King in 2018, Nephew was researching postpartum depression using rodent models. His work quickly shifted to mindfulness and its potential as a non-pharmacological intervention for conditions such as hypertension and chronic pain.

Right away, Nephew said that it was apparent that mindfulness-based stress reduction can have profoundly positive effects on brain structure and function; neuroimaging scans revealed rapid and robust changes that were associated with improved outcomes. “You see physical changes in the brain,” he said. “I was impressed from the start. Pain can change the structure of the brain, and mindfulness can do this as well, but in a beneficial way, helping you to focus less on your pain and more on other positive aspects of your life.” 

Zheyang Wu

Zheyang Wu

Professor, Mathematical Sciences, Biostatistics 

As a form of artificial intelligence, machine learning is nothing without accurate data collection and analysis. Behind every elegant algorithm is a mountain of math and statistics. Wu is making sure the mathematical foundation of the NIH pain grant is sound. He  said the team is developing unique statistical methods to ensure the machine learning being used will predict the likelihood a person will benefit from mindfulness-based stress reduction.

Wu said he is writing a paper on how to calculate sample sizes for machine learning methods. “It’s quite new,” he said.   

He said the NIH pain grant is ambitious, but if he and the team make the right assumptions about their approach to the statistical models underpinning the research, the effort will prove worthwhile.  

Wu said the study is also dynamic; the team will be reassessing and reanalyzing its methods as it continues, making adjustments along the way. “I’m very confident we will bring new knowledge into the field, and advance the field by carrying out this relatively large study using the power of AI.”

Angela Incollingo Rodriguez 

Angela Incollingo Rodriguez

Assistant Professor, Social Science & Policy Studies 

Incollingo Rodriguez is a health psychologist and biopsychosocial researcher whose prior work identifying predictors and mechanisms that drive health behaviors, overall health, and health disparities will help ensure the NIH pain grant team properly collects and analyzes the right data.   

She has extensively studied the relationship between stress and pain, and said the new trial is exciting because it will seek to identify a wide range of factors that could be modifiable, opening up patients to mindfulness-based treatments who might otherwise have been deemed unlikely to benefit. Most studies in this area tend to narrowly focus on a small number of factors, she said.  

 “Pain is an incredibly difficult construct to define,” she said. “We all know what it is, but we struggle to put it into words. The only way to understand it is to get at it from every angle, and this project is truly getting at pain from every angle. We’re looking at all the different modalities that could indicate someone’s feeling pain, and all the different modalities pain could affect: sleep, gait, sensed data.”  

The machine learning aspect of the study may also allow researchers to look deeper into minoritized communities to find out, truly, what ails them.  

Incollingo Rodriguez said the trial’s focus on diverse, under-represented communities could challenge long-held presumptions about who may or may not benefit from mindfulness-based treatments.

For instance, implicit assumptions grounded in structural racism can lead healthcare providers to conclude these types of approaches don’t work for minoritized populations, she said.   

“It’s not because they’re minoritized, it’s because of something else in that predictive picture,” Rodriguez said. “So maybe we can find out from the start that this is actually the barrier – what is the ‘something else,’ and if we can address this piece first, maybe they will get the benefit.”  

For example, if people with high levels of inflammation in a minoritized community are found not to respond well to mindfulness approaches, you can target the inflammation before trying the mindfulness treatment.

The practice could lead to a very personalized approach to behavioral and integrative medicine—"less one-size-fits-all,” Incollingo Rodriguez said, “and more “what-size-fits-who?”

Agu’s expertise in analyzing sensor data using smartphones and fitness trackers will play a critical role in the study. The devices will track several data points, but Agu said of particular interest to researchers will be participants’ circadian rhythms–sleep and wake cycles.

“Sleep has an immense impact on our overall health,” said Agu, who is a co-principal investigator on the study. “An individual in pain is more likely to experience broken sleep, which can lead to a host of other health issues. Mindfulness-based approaches may help participants sleep better, which can reduce some of those other risk factors.” 

The study will include racially and ethnically diverse populations typically underrepresented in both the research and practice of mindfulness-based stress reduction, despite being at increased risk for stress, chronic pain, and the associated adverse outcomes they bring. Participants will be recruited from the Boston metro region through Boston Medical Center and Cambridge Health Alliance, and from the Worcester region through UMass Chan and WPI.  

For physicians, it will be a new day. To be able to predict who would respond well to non-pharmacological interventions will truly save lives. 

Jean King

Peterson Family Dean of Arts and Sciences

Jean King

Partners on the grant and community leaders are excited for the work to begin.  

Dr. Natalia Morone, associate professor of medicine at Boston University Chobanian and Avedisian School of Medicine, a primary care physician at Boston Medical Center, and a co-principal investigator on the study, said the key will be identifying specific markers that indicate people will respond to mindfulness treatment. “We are doing this in an innovative way because we are using machine learning to figure this out” Morone said. “I am very excited to partner with my colleagues at WPI and UMass Chan to accomplish this study. It has the potential to help many people.” 

Dr. David D. McManus, the Richard M. Haidack Professor in Medicine and chair and professor of medicine at UMass Chan, said the medical school will bring invaluable experience to the study gained from overseeing the cores of prominent studies, such as the Framingham Heart Study, National Institutes of Health Rapid Acceleration of Diagnostics (RADx) initiative, and the Risk Underlying Rural Areas Longitudinal (RURAL) study.

“The wealth of knowledge accumulated through the administration and management of critical components in these studies positions us at the forefront of groundbreaking research,” McManus said. “Our enthusiasm is heightened as we join forces with WPI and BU under the capable leadership of Jean King.” 

Dr. Matilde Castiel, commissioner of health and human services in Worcester, said AI is a tool to help the healthcare system deliver better and more personalized care.  

“I am thrilled that WPI will use AI to address chronic back pain and make an impact on the opioid epidemic, which is truly a public health emergency not only in our city and state, but nationally,” Castiel said. “This intervention can decrease the reliance of opioids for chronic back pain and provide a more targeted approach that is specific to the individual.”  