Assessing the antigenicity of tumors is a critical task when designing fine-tuned cancer immunotherapy. Antigen quality, or the affinity of an antigen to a T cell receptor, and antigen quantity, the density of antigen presented on the surface of immune cell targets, both play crucial roles in determining the efficacy of an immune response. Their roles in T cell activation are intimately linked, as low quantities of a high quality antigen and high quantities of a low quality antigen can produce similar immune responses, making it difficult to disentangle the true binding strength of an antigen from the amount of the antigen encountered by a T cell. Because measuring the quality of T cell response to neoantigens is critical to cancer immunotherapy, and because the quantity of an tumor neoantigen is more difficult to measure in vivo and less functionally relevant than its quality, a robust metric for quantifying T cell response to tumor neoantigens should be able to quantify both the quality and quantity of the neoantigen separately. Conventional metrics for assessing neoantigen quality often rely on only a single measurement of T cell activation (e.g. IFN-g ELISPOT), and therefore have difficulties deconvolving antigen quality and quantity. We developed a TECAN robotic platform to track the ex vivo dynamics of T cell differentiation, proliferation and cytokine secretion simultaneously in response to antigens of different binding affinities and concentrations. We introduced a simple neural network to classify “kinetic features” (derivatives/integrals of different observables over different time periods), and deconvolve antigen quality/quantity. We show that this method indeed does allow for more accurate prediction of antigen quality/quantity than any single measurement of an activation marker, and can therefore represent a promising new metric to use for clinical assessment of the immunotherapeutic potential of tumor neoantigens.