Though the trend analysis on data series is very important for the clinical decision making, the objective formulation of the analysis-still remains an inexperienced area. An attempt has been made to develop analyzing methods that would correspond to the sequential changes of clinical data, through an experimental subject of the prognostic estimation of patients with acute myocardial infarction.
Twenty nine cases were divided into two groups, one consisting of 19 who survived a period of two weeks to one year and half after the attack, and the other, who died in the same period. S-GOT, S-LDH, percentage of stab form neutrocytes and blood sedimentation rate were chosen as statistically and clinically significant variables. The analysis was made with two approaches, one being retrospective and the other, prospective.
In the retrospective approach five parameters were extracted so as to represent the characteristics of the data trend pattern. Twenty variables (4 measurements×5 parameters) employed in the discriminant function was effectively reduced to five by an algorithm using the indices of discriminant efficiency and F test.
In the prospective approach, the one-month period after the attack was divided into four intervals by the 4th, 8th and the 16th days. The principal component analysis was first made by contrasting the maximum with the minimum value of measurements in each interval for each patient. A common direction going from the 4th to 2nd quadrant was observed between the survival and dead groups and between the minimum and maximum values. This suggests that the higher measurement values in each interval will indicate the worse prognosis. When the discrimination was made independently for each interval, the probability of miss-classification with F test increased in the order of the 2nd, 3rd, 4th and the 1st intervals.
The retrospective approach is not regarded as favorable because of the delayed decision and the low accuracy of the measurement; the prospective approach is more useful from the view point of clinical practice.
