Abstract
Stroke is the second leading cause of death worldwide and a significant public health concern in Thailand. This article presents predictive factors for in-hospital stroke mortality, utilizing data mining techniques from the Neurological Institute of Thailand’s stroke database. Initially, 41 predictive variables were considered. However, after employing the CFS Subset Evaluator method for variable selection, five key predictive variables emerged: age, first diagnosis (IH, SH), the need for ventilator support, inability to receive a rehabilitation assessment, and the occurrence of pressure sores. Using these selected predictive variables, a classification model was created. The top three classifiers, with the highest F-Measure value of 0.971, were Naïve Bayes, Naïve Bayes Updateable, and Bayesian Network. The knowledge gained from this analysis can be valuable in enhancing the care provided to stroke patients and predicting high-risk stroke-related mortality.
Towanabut, S. ., & Traicharoenwong, C. . (2024). Predictive prognostic factors for stroke mortality in Thailand. Journal of the Thai Medical Informatics Association, 10(1), 8–14. Retrieved from https://he03.tci-thaijo.org/index.php/jtmi/article/view/2701