Recognition of mild cognitive impairment in older adults using a polynomial regression model based on prefrontal...

This study employed a three-minute game-based intelligence test (GBIT) to create a hemoglobin polynomial regression model for early identification of mild cognitive impairment (MCI) in older adults.

210 older adult participants were recruited from community centers in the central region of Taichung City. Working memory (WM) performance in older adults was assessed during GBIT, while hemoglobin responses were measured by near-infrared spectroscopy (NIRS). Variables included oxyhemoglobin (O₂Hb) and deoxyhemoglobin (HHb). Data sequences underwent a fitting procedure using a transformed cubic polynomial function. The transformed coefficients were used as predictors of a logistic regression model to recognize MCI in older adults.

This study confirmed the relationship between age and cognitive performance. The findings demonstrate that the NIRS cubic polynomial function trends during the GBIT test showed significant changes in older adults, increasing with age. Logistic regression analysis identified age and the orientation (coefficient a) of HHb as the main factors for recognizing MCI. The model achieved an overall precision of 83.33 % (sensitivity = 75.00 %; specificity = 84.68 %) with the formula: ln (Odds [MCI]) = -1.64 + 0.57 × HHb_a + 1.40 × age.

NIRS hemoglobin response characteristics during GBIT may serve as an efficient indicator of MCI in older adults. These findings may advance the field of cognitive health evaluation, resulting in earlier detection of cognitive deterioration in older adults.