A method for logical and computational assessment of intended and unintended treatment effects in single cases and small groups is proposed. Insufficient number of observations and not systematically collected information make traditional statistical analysis not feasible, which is a challenge for clinical medicine, drug safety, litigation, etc. A set of assumptions will be formulated. The proposed approach allows for utilizing all available data having a poorly structured information converted to a form feasible for logical and computational analysis. The algorithm operates with a set of categories including a "treatment-outcome complex," "sensitivity to treatment," "spontaneous recovery,""aggregation," etc. The validity criteria will be discussed. The approach can be extended to studying heterogeneity of a population and an individual treatment response.