AI algorithms can assess the pain that someone with sickle cell disease is experiencing by using just their vital signs. Doing so could ensure people receive the most suitable pain management therapy for their condition.
“There’s always a trade-off between giving people sufficient medicine to reduce the pain and giving people too much medication so that they have bad side effects or a higher risk of addiction,” says Daniel Abrams at Northwestern University in Illinois.
But since pain is subjective, it is difficult to measure in a standardised way. Abrams and his colleagues set out to determine whether physiological data that is already routinely taken – including body temperature, heart rate and blood pressure – could be used to devise a system that assesses pain levels in a more objective manner.
The team used data from 46 adults and children with sickle cell disease over a combined total of 105 hospital stays, looking at the physiological data along with patient-reported pain scores to develop models that could deduce pain levels and detect changes in pain level through machine learning.
The researchers then compared their new models against existing ones that try to assess levels of pain but that don’t utilise physiological measurements. The new models outperformed the existing ones.
“The big picture is that we want to better understand how people experience pain,” says Abrams. “We’re hoping that the long-term outcome of this line of research is a more quantitative approach to pain management.”
“I think the most important part of this research is the wider impact that these results could have on pain treatment,” says James Henshaw at the University of Manchester, UK.
This could be especially useful for children, says Abrams, because children often struggle to explain the level of pain they are experiencing.
The team believes that this method can be extended to other types of pain. This study is only the first step in a wider investigation of pain inference and prediction.
“My research group is in the middle of trying to collect a huge set of data on millions of hospitalisations and not just for sickle cell disease but also post-operative pain and other sources of chronic pain,” says Abrams.
Journal reference: PLoS Computational Biology, DOI: 10.1371/journal.pcbi.1008542