The invention relates to the technical field of
pattern recognition, in particular to an AI-based
breathing anomaly recognition method and
system, which comprises the following steps of: extracting a
time sequence analysis trend value, a local amplitude change value and a periodic change value in each group of data based on
breathing frequency
monitoring data and
breathing depth
analysis data; and comparing the
time sequence trend value with the local amplitude change value to obtain a respiratory
rhythm characteristic parameter set, and carrying out classification and partition statistics on the periodic change value to generate a respiratory
rhythm preliminary partition result. According to the method, by carefully analyzing the
respiratory frequency and depth, extracting the trend value, the local amplitude and the periodic change, deeply analyzing the respiratory
rhythm, and accurately capturing the respiratory dynamic state, the
cross matching fluctuation amplitude and the trend rate by utilizing a partitioning and classifying method, the comprehensiveness and the recognition precision in
respiratory monitoring are improved; the abnormal interval deviation value is associated with the
respiratory frequency and depth difference, abnormal points are scientifically marked, and the detection efficiency and accuracy of the respiratory anomaly are effectively improved.