Aim. The objective of the present study was to determine the classification differences in immunological reactivity and to identify its predictors in the newborn infants. Materials and methods. The study involved 115 full-term newborn infants presenting with grade 3 prenatal hypoxic ischemic encephalopathy in the late neonatal period. The features of immunological reactivity under the influence of acupuncture were examined. Statistical processing was carried out by means of discriminant analysis. Results. The assessment and prediction of the effectiveness of acupuncture in the neonates suffering from cerebral ischemia are based on the index of immunological reactivity and the leukocyte index of intoxication, as well as on the ratio of monocytes to band neutrophils content. For generation of the group classifier of immunological predictors in a newborn infant and development of indications for reflex therapy, nine parameters of interest were measured. The group specificity of the child was determined by three variables, viz. leukocyte index of intoxication, monocyte and band neutrophil counts with values of the Fisher's exact test (F) and reliability (Wilks Lambda 0.90894; approximation F (3.144) =4.809; p<0.0032). The partial Wilks Lambda values showed that the greatest contribution was provided by the leukocyte index of intoxication and monocytes. Prediction accuracy of the classification matrix in the standard treatment group reached 30.8% and 91.7% respectively when reflex therapy was included in the combined rehabilitation treatment. Overall, classification accuracy amounted to 70.3%. The presence of distinctive changes in the subgroups preconditioned a personalized approach to the prescription of reflex therapy to the newborn infants and the choice of the treatment modality on an individual basis (parent, child, or both) in the "mother-newborn" system. The variant of treatment was determined by comparing the values of the results of the formulas. The newborns were referred to the subgroup with the highest value of the classification function. The predictors made it possible to reliably distinguished the second (p=0.032) and the third (p=0.022) subgroups from the first one, with some degree of overlapping between the edge zones of centroids of the second and third subgroups (p=0.073). Therefore, the sensitivity of classification in the individual subgroups was lower than in the group model and was estimated at 34.4, 71.9, and 65.6%for the first, second and third groups, respectively. Conclusion. The mathematical models can discriminatеbetween the immunological characteristics and predict them in individual newborn infants; also, they can be helpful for preventing the disruption of their adaptation process.