Abstract
The final color of layered resin composite restorations is influenced by complex optical interactions between composite layers and surrounding backgrounds. Therefore, this study investigated whether background-dependent optical measurements and translucency parameters could be used to predict final color coordinates and assessed model performance in an independent composite system. During model development, five composite shades (D3, D2, D1, E2, and E1) were used in both the 2-mm base and 1-mm top positions, generating all 25 possible ordered bilayer combinations. Background-dependent L*, a*, and b* coordinates and translucency parameters obtained from independent 1-mm and 2-mm layers were used to develop multivariable linear regression models for predicting the color coordinates of 3-mm layered specimens. Model performance was assessed using adjusted R², mean absolute error, root mean square error, and leave-one-out cross-validation. Independent-system evaluation was performed using 16 layered combinations from one additional composite system, and predictive accuracy was evaluated using CIEDE2000 color differences (ΔE00). The regression models demonstrated adjusted R² values of 0.799, 0.843, and 0.964 for the L*, a*, and b* coordinates, respectively. During the independent-system evaluation, the mean ΔE00 value was 0.939 ± 0.333; 37.5% of the combinations were at or below the perceptibility threshold (ΔE00 ≤ 0.8), while all combinations (100%) were within the clinical acceptability threshold (ΔE00 ≤ 1.8). Background-dependent optical measurements and translucency parameters enabled accurate and clinically acceptable estimation of the final color coordinates of layered resin composites. The low prediction errors observed in one additional composite system provide preliminary evidence of predictive performance beyond the development material under standardized experimental conditions.