Yang M C, Marks R G, Clark W B, Magnusson I
Department of Statistics, University of Florida, Gainesville 32611.
Stat Med. 1991 Jul;10(7):1089-97. doi: 10.1002/sim.4780100709.
Three methods for detecting the progression of periodontal disease based on the change in attachment level on teeth are compared from a statistical point of view. Regardless of gradual or burst disease progression theories, the regression method is found to be the most efficient way to detect changes. The exact false alarm rate is computed when the decision is based on the maximum regression coefficient at all sites during several visits.
从统计学角度比较了基于牙齿附着水平变化检测牙周疾病进展的三种方法。无论疾病进展是渐进性还是突发性理论，回归方法被发现是检测变化的最有效方法。当决策基于多次就诊期间所有部位的最大回归系数时，计算出确切的误报率。
