The authors of this paper applied a new approach combining text mining and principal component analysis (PCA) to objectively determine the actual state of regional COVID-19 strategy meetings and verified its utility. The authors used text mining to analyze meeting minutes and extracted words with high phase ubiquity by co-occurrence analysis. Then, they selected words symbolizing the meeting contents ("report," "prevention," "rules," and "decision") and performed PCA using the occurrence rates

more »

... f these words as variables. Two principal components (PC1, PC2) were set. For PC1, we observed maximum factor loading for "decision" (0.81) and minimum for "report" (-0.72), so we considered this axis to show the "depth of meeting discussions." For PC2, we observed maximum factor loading for "prevention" (0.81) and minimum for "rule" (-0.76). We considered this axis to show "regional infection status." When we created a plot of all 44 meetings, Phase 1 occurred in quadrants 3 to 4 (knowledge sharing), phase 2 began in quadrant 1 (preparation for spread), and phase 3 shifted to quadrant 2 (response to spread) with significant differences between these phases. Our findings suggest that the actual state of regional COVID-19 strategy meetings could be objectively determined by using a combination of text mining and PCA.