| abstract |
Systems, methods and computer-readable media are provided for identifying persons who are likely to benefit from a neurosurgical procedure, such as lobectomy, hemispherectomy, lesionectomy, callosotomy, and the like. Measured values of physiological variables may be combined via a multi-variable predictive model. In some embodiments, this may take the form of a multinomial logistic regression classifier. In other embodiments, the evidence-combining may be implemented via a neural network or support vector machine or similar machine learning or classification methods. In an embodiment, a leading indicator of near-term responsiveness to the regimen may be provided to a caregiver, such as a neurologist, and further may be integrated with case-management software and an electronic health record decision-support system. |