Immune-targeted therapies that activate effector lymphocytes such as Natural Killer (NK) cells are currently being investigated for the treatment of Multiple myeloma (MM), the second most common form of hematological cancer. However, individual NK cells are highly heterogeneous in their cytolytic potential, making it difficult to detect, quantify and correlate the outcome of dynamic effector-target cell interactions at single cell resolution. Here, we present a microfluidic bioassay platform capable of activity-based screening of cellular and molecular immunotherapies. We identified distinct functional signatures associated with NK-MM cell interaction. The addition of immunomodulatory drug lenalidomide altered responses of NK-susceptible MM cells but not that of NK-tolerant MM cells. Antitumor cytotoxicity was significantly increased by the blockade of PD1/PDL1 axis as well as the clinically relevant cell line NK92, which were used to construct molecular logic functions (AND and NOT gates). A predictive agent-based mathematical model was developed to simulate progressive disease states and drug efficacy. The findings of the current study validate the applicability of this microfluidic cytotoxicity assay for immunotherapy screening, biocomputation and for future employment in detection of patient-specific cell response for precision medicine.