Back

"Devil's stairs", Poisson's Statistics, and Patient Sorting via Variabilities for Oxygenation: All from Arterial Blood Gas Data

Chuiko, G. P.; Darnapuk, Y. S.; Dvornik, O. V.; Krainyk, Y. M.; Yaremchuk, O. M.; Diadenko, A.

2021-08-13 bioinformatics
10.1101/2021.08.10.455835 bioRxiv
Show abstract

This report deals with arterial oxygen saturation (SaO2) for healthy adults. A comparably small data set (20 persons) holds 3-minute records of SaO2. The sample rate was 200 Hz. The charts have the looks of a "devils stairs." A few (from 1 to 10) detectable oxygenation levels form the stairs treads, more or less long. "The risers" have two types (up and down), and all have virtually the same height, about 1 %. The inter-level shifts ( 0 to 42 switches per record) turned out a rare event at the actual sample rate. The number of switchings meets the Poisson distribution. There were found three visibly varied intensities for the switch-overs within the data set. Histograms also show the co-existing of no fewer than three subsets into the data set. The subsets differ by the intensity of switch-overs, amounts of possible levels, relative frequencies of most probable levels (modes), etcetera. In short, those all are diverse variability quantifiers. The higher variability subset has about 25 %, the lower one - 45%.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.