Introduction of turbidimetric homogeneous immunoassays made the determination of plasma ferritin concentration wide-ranging available. However, high-dose hook effect or prozone effect occurring at samples with high ferritin concentration can lead to false-negative results. According to the authors, this phenomenon has considerable clinical significance, in patients with iron-overload disorders false-negative laboratory values may result in inaccurate diagnosis. The prozone effect can be eliminated by reaction kinetic analysis of measurements. The authors developed a neural network classification procedure based on artificial intelligence technology for the recognition of the reactions with differing kinetic flow, and made a computer software for helping the application of the classification system. False-negative results can be filtered using this new technology following the laboratory determination, thus sensitivity of plasma ferritin determination may become safe enough even in case of high concentration samples.