Machine Learning for Chronic Pain Management Optimization Within a Health Equity Framework

One in five U.S. adults (50 million) has chronic pain. Chronic pain can result in lower quality of life, higher risk of disability, and higher risk of military separation. Chronic pain is also associated with an increased risk of suicide, but this risk can be lowered if patients receive targeted pain management therapies. Over the past decade, the Department of Defense has invested a lot of effort and funding to improve pain management for active-duty Service Members, military retirees, and their family members. Some of these efforts include making complementary and integrative health therapies more available. This includes physical therapy, behavioral health (psychological) therapy, acupuncture, and movement therapies like yoga and Tai Chi. Other efforts focus on changing healthcare provider practices, like lowering opioid prescription doses and increasing patient referrals to complementary and integrative health therapies.

Previous studies from randomized control trials show that some complementary and integrative therapies work, but only for some health conditions or outcomes. Because these trials are not like real-world clinical care, healthcare systems must analyze the effectiveness of these therapies using their real-world clinical data. In real-world studies, such as studies from the Veterans Administration and civilian healthcare systems, some patients do not have equal access to the same therapies as other patients. As a result, many patients have worse outcomes than others. Like the Veterans Administration, it is important for researchers to examine real-world, big-data in the Military Health System to make sure all patients have equal access to treatments and benefit equally from these treatments. This type of big-data study has not been done in the Military Health System. Therefore, this study will determine who is receiving these therapies, when they are receiving them, why they are receiving them, and how well they work (and for which health conditions) across many outcomes.

The information from this study will be helpful in many ways. Because the Military Health System is large and expands across the globe, these analyses will provide military leaders with the information they need to ensure all patients have equal access to and benefit from these therapies and support program implementation efforts. When big-data (machine-learning) studies and provider tools do not account for health inequities, the tools can make treatment access and outcomes worse for patients from underrepresented and marginalized groups. Because the analyses are first focusing on equality in therapy access and outcomes, then accounting for these differences in treatment comparisons, the results can be used to make treatment access and outcomes better for more patients. For example, the study methods and results can be transformed into provider dashboards that display treatment recommendations, based on the patient's symptoms and health conditions, as well as the treatments that are locally available. For this study, the research team is building a blueprint so that other military healthcare stakeholders can also apply this big-data approach to other healthcare fields, like diabetes management and behavioral medicine. In this way, the results of this study will continue to benefit more and more patients in the Military Health System.

This study includes experts from a variety of different scientific areas, including pain management, primary care and family medicine, military healthcare, big-data analyses, healthcare disparities and equity, and decision support tool development. The study is using healthcare data that has already been collected, and the study team completed a pilot study to make sure this type of study was possible. The study team has set up the analyses so that the first set of results about pain management therapy access will be available within 1 year. In the second year, they will have information about pain management therapy outcomes. In the third year, they will have additional information about which therapies or treatments work best for patients, depending on their health conditions, as well as how provider referral practices can make pain management more successful.

With the results from this study, the negative impact of chronic pain in the Military Health System can be lowered, which would then lower the negative impact of chronic pain in the Veterans Administration. This may mean that fewer Service Members separate from the military because of chronic pain. More recently, the Defense Health Agency has announced plans to reduce healthcare access for retired Service Members and family members of active-duty and retired Service Members at many military treatment facilities. However, this may mean that more patients with chronic pain will not be able to access complementary and integrative healthcare, especially if retirees and family members are located in rural areas and areas without many healthcare providers. The results of this study will estimate what would happen if leadership were to carry out these plans – and the results will be shared with leadership to support any changes in plans and policies.