Improvement in the information content of the Glasgow Coma Scale for the prediction of full cognitive recovery after head injury using fuzzy logic

Background. The objective of this study was to modify the existing Glasgow Coma Scale (GCS) into a fuzzy GCS by using fuzzy information representation and fuzzy inferencing. The study compared the information content of the existing GCS with the new fuzzy GCS for prediction of full cognitive recovery in patients with head injury. Methods. A record-based study was conducted at the Government Medical College and Hospital, a tertiary care facility In Nagpur, India. The study, which covered the period from January 1 to December 31, 1997, included 253 patients with head injuries. Opinions of 17 clinical experts who routinely deal with head injury cases were used for the construction of the fuzzy GCS. Results. By using the max operator for summarization, eye, motor, and verbal stimuli were all significantly associated with the possibility of full cognitive recovery with the fuzzy GCS (P < .001). Nonspecificity of the classical GCS, the min-operated fuzzy GCS, and the max-operated fuzzy GCS was comparable. A reduction in Shannon entropy was maximum with the max-operated fuzzy GCS. Mini-operated fuzzy GCS better predicted a lack of full cognitive recovery. Conclusions. Fuzzy GCS substantially improves the information content for prediction of the possibility of full cognitive recovery after head injury. Eye, motor; and verbal stimuli all uniquely and significantly contribute to prediction of this possibility. We recommend the use of fuzzy GCS for prediction of the possibility of full cognitive recovery in patients with head injuries.