Introduction: BACE-1 is an aspartate protease that is responsible for the proteolysis of amyloid precursor proteins into beta-amyloid (Aβ), a neurotoxic peptide in patients with Alzheimer’s disease (AD). As such, BACE-1 is a prime pharmacological target in the control of Aβ in the brain and its inhibition will be a sound approach in AD therapy. The majority of BACE 1 inhibitors in past and current clinical trials do not show significant selectivity for BACE 1 over BACE 2. Several BACE 1 inhibitors have been progressed into clinical trials, all of which unfortunately have been terminated due to toxicity and lack of efficacy. These limitations make that the β-Secretase (BACE 1 and BACE 2) enzimes could be two attractives drug target for the development of news BACE 1 selectivity inhibitors over BACE 2 in the AD. Methods: Quantitative structure activity relationship (QSAR) models were used for identification of selective potential inhibitors of BACE 1 protein. A dataset of 76 inhibitors collected from the literature with their IC50 being determined experimentally has been assembled. This dataset contains diverse compounds with different scaffolds and physical-chemical properties, covering a wide chemical space in the drug-like range. The alvaDesc 2.0.6 software was used with a series of molecular descriptors. The prediction of biological activities for a data set of 157 modified ribonucleotides is also presented. Results: The QSAR model for BACE 1 estimated that molecules with the following properties showed a positive contribution in the inhibitors activity: low value of frequency C-C at topological distance 3 and high values of molar refractivity, topological polar surface area using N, O, S, P polar contributions, octanol-water partition coefficient and a lead-like score drug-like indices. Besides, the QSAR model for the BACE 2 estimated that molecules with other properties as described below also increases the BACE 2 potency: a low value of sum of conventional bond orders (H-depleted), Quantitative Estimation of Drug-likeness (unweighted), high values of distance/detour ring index of order 6, and modified lead-like score (6 rules). Conclusions: The combined possibilities of quantum and physicochemical molecular descriptors, together with the Machine Learning techniques, allowed us to generate QSAR models, capable of discriminating between BACE 1 over BACE 2 inhibitors. QSAR models allowed the identification of five modified nucleotides as selective inhibitors against BACE 1 over BACE 2 in Alzheimer’s disease. Phosphate group in the selective modified nucleotides have a positive impact in the inhibitor’s activity. According the value of the significant molecular descriptors include in the QSAR models of BACE 1, a phosphate group in the molecule 156, the most active modificated ribonucleotide increases the BACE 1 inhibitor activity.