| 摘要 |
<P>PROBLEM TO BE SOLVED: To diagnose an object by early and surely detecting a variation in state occurring in the cardiogram of a patient from characteristic amounts repeatedly appearing in the electrocardiogram, for example, from an interval between R and R. <P>SOLUTION: Normally, a distribution calculation means 3 calculates the characteristic amounts repeatedly appearing in the electrocardiogram as time-series data by using electrocardiogram data measured in an period of INT1 as an input, and generates the set C of a difference between adjacent elements of the time-series data. Also, a processing for randomly extracting n elements from the set C is repeated m times to provide m subsets. The average value μ and a standard deviation σ of a normal distribution when the set of the average value of each of the m subsets is regarded as the normal distribution, and the calculated value is taken as a normal distribution presenting the normal state of a variation in characteristic amounts. A state detection part 4 provides a set CT for diagnosis in the same manner as in the set C from the electrocardiogram data in a period shorter than the period INT1, and a diagnosing part 5 diagnoses by comparing a difference between the average value of the set CT and the average value of the normal distribution with the standard deviation σ. <P>COPYRIGHT: (C)2003,JPO |
