診断試験 2×2 表計算機
本ツールは臨床研究・統計の教育および探索用です。入力に基づきブラウザ内で計算し、データはサーバーに送信されません。結果の解釈は原典の手法・日本の関連指針および施設プロトコルに従ってください。診断や治療、論文投稿の最終判断の代替ではありません。
① Fill in the 2×2 table
| ゴールドスタンダード + | ゴールドスタンダード − | |
|---|---|---|
| 検査 + | ||
| 検査 − |
使い方
2×2 指標の意味は？
感度 = TP/(TP+FN), reflecting few missed cases; specificity = TN/(TN+FP), reflecting few false alarms; PPV is the proportion of test-positives who are truly diseased; NPV is the proportion of test-negatives who are truly non-diseased.
的中率が変わるのはなぜですか？
PPV and NPV depend on the population prevalence: for the same test, PPV is higher in a high-prevalence population and lower in a low-prevalence one. When citing literature predictive values, check whether their prevalence matches your population; likelihood ratios are unaffected by prevalence and more stable.
尤度比はどう読みますか？
The larger the positive LR (+LR), the more a positive test indicates disease; above 10 is usually strong diagnostic value. The smaller the negative LR (−LR), the better a negative test rules disease out; below 0.1 is strong. The diagnostic odds ratio (DOR) summarizes overall discrimination.
信頼区間にWilson法を使うのはなぜですか？
For proportions, the traditional Wald interval is distorted at small samples or when the proportion is near 0/1. The Wilson method is more robust at small samples and never goes out of bounds, so sensitivity, specificity and other proportions here use Wilson 95% CIs.