summary: New research shows that routine hospital blood tests can help predict the severity of spinal cord injuries and the likelihood of survival. Researchers used machine learning to analyze data from thousands of patients, and patterns of blood markers, such as electrolytes and immune cells, predicted recovery outcomes 1-3 days after admission.
Unlike neurological testing, which depends on a patient's reactivity, this method provides objective and reliable insights. The findings may improve emergency care and resource allocation for spinal cord injuries worldwide.
Important facts
- Predicted value: Blood marker patterns predict the severity of the injury and mortality.
- AI Insights: Machine learning has improved accuracy with more test data over time.
- Practical: Routine testing is affordable, available in all hospitals and is more accessible than MRI and advanced biomarkers.
sauce: University of Waterloo
According to a study from the University of Waterloo, daily blood samples that are ingested daily at any hospital and are tracked over time may help predict the severity of the injury and even provide insight into mortality after spinal cord injury.
Using advanced analytics and machine learning, a type of artificial intelligence, the research team evaluated whether routine blood tests serve as early warning signs of outcomes in patients with spinal cord injury.
According to the World Health Organization, in 2019, more than 20 million people were affected by spinal cord injuries, with 930,000 new cases each year. Traumatic spinal cord injuries often require intensive care, and a variety of clinical symptoms and recovery trajectories are characterized, particularly in the emergency department and intensive care unit, complicating diagnosis and prognosis.
“Regular blood tests can provide doctors with important and affordable information that will help them predict the risk of death, the presence of injuries and how serious it is,” said Dr. Abel Torres Espin, a professor at the School of Public Health Sciences in Waterloo.
The researchers sampled hospital data from more than 2,600 patients in the United States. They used machine learning to analyze millions of data points and discovered hidden patterns of common blood measurements, such as electrolytes and immune cells ingested in the first three weeks after spinal cord injury.
They found that these patterns can help predict recovery and injury severity without early neurological examinations, which are not always reliable, as they depend on patient reactivity.
“While a single biomarker measured at a single time point can have predictive power, the broader story lies in the changes that are shown over time with multiple biomarkers,” said Dr. Margie Mussabi Rij, a postdoctoral scholar at Torres Espin's lab in Waterloo.
Models that did not rely on early neurological assessments were accurate in predicting mortality and injury severity 1-3 days after hospital admission, compared to standard non-specific severity measures often performed on the first day of arrival in the intensive care unit.
The survey also found that accuracy increases over time as blood tests increase. Other measurements such as MRI and Fluid Omics-based biomarkers can also provide objective data, but they are not always readily accessible across healthcare settings. On the other hand, daily blood tests are economical, easy to obtain and are available in all hospitals.
“Predicting the severity of injuries over the first few days is clinically relevant to decision-making, but neurological assessment alone is a challenging task,” Torres Espin said. “It shows the possibility of predicting whether the injury is completely athletic or incomplete. It is routine blood data and incomplete early after an injury, indicating an increase in predicted performance over time.
“This basic work opens new possibilities in clinical practice and allows for more detailed decisions regarding treatment priorities and resource allocation in the critical care settings for many physical injuries.”
About this spinal cord injury and neurology research news
author: Lyon Jones
sauce: University of Waterloo
contact: Ryon Jones – University of Waterloo
image: This image is credited to Neuroscience News
Original research: Open access.
“Modeling the trajectory of routine blood tests as a dynamic biomarker of outcomes in spinal cord injury,” Abel Torres Espínetal. NPJ Digital Medicine
Abstract
Modeling trajectories of routine blood tests as dynamic biomarkers of spinal cord injury outcomes
Blood tests collected routinely can reflect underlying pathophysiological processes.
The dynamics of routinely collected blood tests demonstrate that they retain the validity of predictions of acute spinal cord injury (SCI). Using imitation data (n= 2615) Modeling and Track SCI Research Data (n= 137) For validation, multiple trajectories of common blood markers were identified.
We developed a machine learning model for dynamic prediction of in-hospital mortality, SCI occurrence in spinal trauma patients, and SCI severity (movement complete and incomplete).
In-hospital mortality model achieved an off-train ROC-AUC of 0.79 [0.77–0.81] Improved to 0.89 on the first day after the review [0.88–0.89] By day 21, the highest ROC-AUC was 0.71 to detect the presence of SCI after spinal trauma [0.69–0.72] Achieved by the 21st day. By day 7, the ROC-AUC for SCI severity was 0.81 [0.77–0.85].
Our complete model outperformed the severity score SAPS II after seven days of hospitalization.
