| pubmed-article:8026251 | pubmed:abstractText | Continuous 24-hr intragastric pH-metry was prospectively performed in 801 subjects with different clinical conditions using two pH electrodes placed closely adjacent. The aim was to assess the in situ repeatability of the test and to verify whether the removal of artifacts, interference, and noise usually superimposed onto the fundamental signal recorded by the measuring apparatus improves the clinical usefulness of experimental data. The following debugging/filtering procedure was used: first, pH recordings of each channel were amended separately from artifacts, then they underwent 7 min windowed median interference debugging, and finally Wiener noise filtering was applied. Afterwards, the 24-hr mean pH profile was obtained in each subject by averaging the pH tracings of the two channels every minute (1440 data points/24 hr). The efficiency of this procedure was assessed at each step by evaluating the difference among groups using the O'Brien test, a distribution-free nonparametric method well-suited for evaluating differences among groups allocated onto a two-way layout. The differences among groups calculated from raw pH data of the single channels can be very misleading, in that it is possible to find that they are significant on one channel and not significant on the other channel. Conversely, the significance of the differences among groups increases progressively at each step of the above debugging/filtering procedure applied to raw pH profiles of each channel. Seven minutes was shown to be the most suitable time lag for windowed median removal of interference.(ABSTRACT TRUNCATED AT 250 WORDS) | lld:pubmed |
