Cognitive ability screening instrument (CASI) is a cross-cultural mental screening test to evaluate the severity of dementia of Alzheimer's disease. In this research, a characteristic-point-based fuzzy inference system (CPFIS) is proposed to predict the CASI scores from the single-photon emission computed tomography (SPECT) volumes. Experiment results showed that CPFIS could provide a rough estimation of CASI scores that the P value for the hypothesis that the average of absolute errors of CPFIS is less than the average of absolute differences from the mean values was 0.0447. For comparison, a weighted support vector regression (SVR) machines was also tried. The performance of CPFIS was slightly better than SVR in this study (P=0.068).