Recently development of computer-aided herbal plant extraction process design has received attention for highly positive future outlook of global herbal industries. Solid-liquid equilibrium (SLE) prediction models comprise a core part of the chemical engineering knowledge base required for the computer aided process design. In this study, solubility model was developed to predict the solubility of several Orthosiphon stamineus (Misai Kucing) phytochemicals (i.e. Sinensetin (SEN), Eupatorin (EUP) and Tet-ramethylscutellarein (TMF)) in seven solvents, in addition to creating physicochemical property database comprising of melting point temperature, fusion enthalpy, and SLE solubility data. Quantitative Structure-Property Relationship (QSPR) model was developed for fusion enthalpy prediction with absolute average relative deviation (AARD) of 933%. For improvement of KT-NIST-UNIFAC model, new subgroups (i.e. aC - CO (fused), aC - O (fused) and C=C-cyclic - OH) were introduced to better represent the molecular structure of studied flavonoid compounds, and binary interaction parameters were regressed using compiled SLE solubility data. Regressed KT-NIST-UNIFAC model exhibited better performance with AARD of 5.27%. Subsequent simulation study using improved model indicated that acetone was compatible for SEN, EUP, and TMF extraction. (C) 2016 Elsevier B.V. All rights reserved.