Piecing Together the Puzzle of Chronic Low Back Pain

Modeling pain

From a plastic replica of the spine to a complex 3D one, a model is a representation of something from the real world, explained Grace C.Y. Peng, Ph.D., a modeling expert at the National Institute of Biomedical Imaging and Bioengineering at NIH. Peng participates in BACPAC’s Theoretical Models for Chronic Low Back Pain working group.

“We like to use models to try to understand the system as much as we can,” Peng said.

With a good model, researchers can simulate injury on a person’s back or cause anxiety and see how that changes other variables, such as the pain experience — if the model has been checked against enough data from people with pain. Or a researcher might use the model to see how different treatments might alter the pain.

The BACPAC team will develop an integrated model that shows how the back, the social environment, and the brain work together to produce the experience of pain.

Getting started

Several steps contributed to the model. The working group started with a sketch showing how different factors interact to cause pain — the simplest version of a model.

This initial diagram brought together multiple ideas about the causes of chronic low back pain. The sketch included about two dozen boxes with labels like “Physical Activity” and “Inflammation.” Arrows showed which boxes influence each other.

For example, an arrow connecting “Pain Experience” to “Physical Activity” was labeled “Pain Avoidance.” This represents how people with chronic pain often stop exercising or change the way they move because they want to avoid causing more pain. However, avoiding physical activity can lead to getting out of shape and to even more pain — which is also reflected in the simple model, through a series of other boxes and arrows showing the chain of links between physical activity and pain.

Refining the model

The basic diagram provides a framework for input from additional BACPAC researchers.

Refining the model includes more detail for each element — for example, defining the missing data on the anatomy of the back, or on pain beliefs. The research consortium also must agree on standard ways to make measurements, so that data from multiple studies can be combined.

Another part of refining the model will be finding computer-friendly ways to describe the relationship between different elements. For some parts of the model, mathematical equations will work. For example, math can describe the movements and connections of the vertebrae, discs, and tendons. For others, the researchers may use an artificial intelligence technique called machine learning, in which a computer looks for patterns in existing data, such as electronic health records or medical images.

When the BACPAC model is more fleshed out, researchers will be able to use it to inspire new studies about back pain.

“The model is a tool,” Peng said. “It’s a platform for creating new hypotheses, and it’s something people should be continuously using to map out the knowledge we know as well as identify what we don’t know.”

For example, through making the model, the group might realize that not enough is known about the meaning of a particular result on an MRI. Researchers might try to learn more by digging through data from electronic health records, or they might design a new study.

Feeding in data

The model is only one part of BACPAC’s collective effort. The consortium is also carrying out many studies on back pain. The findings from these studies will help add detail and information to the model.

Seven Technology Research Sites will conduct studies to develop new technologies focused on back pain, such as better tools for diagnosis or better treatments for chronic low back pain.

Three Mechanistic Research Centers will sign up thousands of people with back pain and take “as many measurements as we can, within reason,” Lotz said.

In addition, two institutions will carry out studies with humans to test new treatments — including alternative medicine, nonaddictive drugs, and devices — for back pain.

The data generated by the studies will be fed into the model to add to the understanding of chronic low back pain. As new information is added to the model, researchers could use the model to explore new connections. Perhaps the model will reveal a new contributor to pain or show a surprising way that people with chronic low back pain differ from people without back pain. It could even point the way to new treatments for chronic low back pain.

“I think both physicians and patients would love to have something that’s more holistic and data-driven to help guide them,” Lotz said.

The ultimate goal for BACPAC is for the model to be turned into a tool that health care practitioners can use. Someday, a person whose back has been hurting for a long time might go to the doctor, fill out a questionnaire, then have a diagnostic test or two. The BACPAC model could take in all the data and then suggest a personalized treatment plan based on the specifics of this person’s life and conditions. For example, it might estimate the probability that a person would benefit from physical therapy, psychotherapy, or surgery. It might even suggest the type of therapy, the dosage of a drug, or a surgical procedure that would be likely relieve the back pain. The research and modeling efforts in BACPAC represent progress toward a future of medicine in which all pain can be treated in a way that is personalized to the patient.