The research scientist from the University of Texas Medical School discussed how machine learning was most effective in predicting delayed cerebral ischemia when it took human-derived variables into account.
“When we combined [human intuition and machine learning], we were able to get better scores. That suggests that, in fact, humans and computer models have to work together instead of trying to replace one with another.”
Jude Savarraj, PhD, research scientist, department of neurosurgery, University of Texas Medical School at Houston, and colleagues recently published a study that suggests machine learning (ML) models used to predict delayed cerebral ischemia (DCI) and functional outcomes significantly outperformed standard models (SMs) when used in subarachnoid hemorrhage (SAH) care.
The ML model outperformed the SM in area under the curve (AUC) by 0.20 (95% CI; -0.02 to 0.4) for DCI, 0.07 (standard deviation [SD], 0.03; 95% CI, 0.0018–0.14) for discharge outcomes, and by 0.14 (95% CI, 0.03–0.24) for 3-month outcomes. Additionally, ML models matched physician performance in predicting 3-month outcomes.
The ML models initially only used routine variables included in a patients’ electronic medical record (EMR), but Savarraj and colleagues found that the ML models performed even better when they also assessed clinician-determined Hunt-Hess Scale score in addition to standard EMR variables (AUC, 0.85; SD, 0.05; 95% CI, 0.75–0.92). Only using EMR data yielded an AUC of 0.81 (SD, 0.05; 95% CI, 0.71–0.89; P <.05).
In this interview, NeurologyLive spoke with Savarraj to learn more about the use of machine learning and synergistic opportunities to combine human and computer skill.