Prototype algorithm for automated determination of gastric slow wave characteristics.

An algorithm for determining the frequency and propagation time of the gastric slow wave has been designed for integration into a demand gastric pacing system. The algorithm analyses the serosal activity in both the time and frequency domains, and the results are compared to produce a conclusion only when the values are within 5% of each other. Thus, the probability of inappropriate intervention is reduced, at the expense of unidentified segments. The system is verified by comparing the conclusions produced by the algorithm with conclusions from hand analysis of seven canine and one human serosal recordings. The algorithm correctly identifies the slow-wave frequency in the distal portion of the stomach for 90% of the segments, while producing no incorrect results. Slow-wave propagation times in the antrum are correctly identified for 84% of the segments, with no incorrect identifications.