Diabetes mellitus is a common chronic metabolic disease that constitutes a risk factor to patient infected by Covid-19. Aldose reductase enzyme catalyzes the metabolic glucose-sorbitol conversion for the polyol pathway in diabetic conditions. The accumulated sorbitol in the tissues has been reported to be responsible for diabetic complications such as cataracts, retinopathy, neuropathy, and nephropathy. Inhibitors of aldose reductase are considered as therapeutic target in prophylaxis and treatment of these affections. A particular case, have been studies the flavonoids isolated from fruits, vegetables and medicinal plants. In the current paper we development a Generalized Linear QSAR model using MATLAB and molecular descriptor implemented in DRAGON software. A result shows a model adjusted whit R² of 0.948. The model was extensively validated according to OECD regulatory principles by mean of internal and external validation exercises. In addition, the applicability domain was obtained to warranty the trustworthy of the predictions. Due to it predictive power (R²ext = 0,943) the model was used to predict the ALR2 inhibition by the flavonoids reported in Jatopha gossypifolia L. The ellagic acid was the most promising metabolite (pIC₅₀ predicted = 12.69), which is into the applicability domain and have drug-like properties for oral administration. Finally, we can conclude that proposed tools are useful to the quick and economic identification of drug potential candidates against ALR2 in diabetic complications.