VIDEO: Unsupervised machine learning may be able to classify synovial fluid samples
Key takeaways:
- Results showed an unsupervised machine learning model may be able to classify synovial fluid samples by infection status.
- The model analyzed synovial fluid tests and matched these into disease-relevant clusters.
In this video from the Musculoskeletal Infection Society Annual Meeting, Carl A. Deirmengian, MD, discussed results that showed unsupervised machine learning may be able to accurately classify synovial fluid samples by infection status.
“If we are able to diagnose infection with a machine learning model based on synovial fluid lab data, that would democratize [periprosthetic joint infection] PJI diagnosis because a laboratory could accomplish this,” Deirmengian said. “If there is this underlying natural clustering of samples, a laboratory could take all of the synovial fluid data, combine it into the AI model and gain a diagnosis from that multidimensional information.”
He added, “That diagnosis would be available to the doctor from the laboratory, which would bring the skill of all of the experts in the field to everyone across the country who is sending in lab data.”
Collapse
Deirmengian CA, et al. Paper 1295. Presented at: Musculoskeletal Infection Society Annual Meeting; Aug. 2-3, 2024; Durham, North Carolina (hybrid meeting).