Logistic regression exploration of host–dopant compatibility in MR-TADF-based OLEDs

The process of selecting an optimal host material for organic light-emitting diodes (OLEDs) based on multiple resonance thermally activated delayed fluorescence (MR-TADF) materials has remained time-consuming and resource-intensive. To accelerate this process, we utilize the MR-TADF material DtCzB-mDS from our previous work as the dopant, pairing it with 14 commonly used host materials from the literature to fabricate OLEDs for performance evaluation. Subsequently, we apply a logistic regression algorithm from machine learning to derive empirical formulas linking the host–dopant emissive system to key device performance metrics. The analysis of these formulas reveals correlations among the variables; for instance, the highest occupied molecular orbitals (HOMO) difference between the host and dopant shows a strong negative correlation with the highest external quantum efficiency (EQE max ) and luminance, with correlation coefficients of −0.67 and −0.54, respectively. Overall, this work offers valuable insights into optimizing host material selection for enhanced OLED performance.