Prediction of cellular response to multiple stimuli is central to evaluating patient-specific clinical status and to basic understanding of cell biology. Cross-talk between signaling pathways cannot be predicted by studying them in isolation and the combinatorial complexity of multiple agonists acting together prohibits an exhaustive exploration of the complete experimental space. Here we describe pairwise agonist scanning (PAS), a strategy that trains a neural network model based on measurements of cellular responses to individual and all pairwise combinations of input signals. We apply PAS to predict calcium signaling responses of human platelets in EDTA-treated plasma to six different agonists (ADP, convulxin, U46619, SFLLRN, AYPGKF and PGE(2)) at three concentrations (0.1, 1 and 10 x EC(50)). The model predicted responses to sequentially added agonists, to ternary combinations of agonists and to 45 different combinations of four to six agonists (R = 0.88). Furthermore, we use PAS to distinguish between the phenotypic responses of platelets from ten donors. Training neural networks with pairs of stimuli across the dose-response regime represents an efficient approach for predicting complex signal integration in a patient-specific disease milieu.