Table 2.

Principle of disproportionality measure and standard of signal detection.

Methods	Calculation formula	Inclusion standard of positive signal
ROR	$\mathrm{ROR}=\frac{(a/c)}{(b/d)}$	a ⩾ 3 and 95% CI > 1
	$\mathrm{SE}(\ln \mathrm{ROR})=\sqrt{\left(\frac{1}{a}+\frac{1}{b}+\frac{1}{c}+\frac{1}{d}\right)}$
	$95\%\mathrm{CI}=e^{\ln (\mathrm{ROR})\pm 1.96\sqrt{\left(\frac{1}{a}+\frac{1}{b}+\frac{1}{c}+\frac{1}{d}\right)}}$
	$\mathrm{PRR}=\frac{a/(a+b)}{c/(c+d)}$
PRR	$\mathrm{SE}(\ln \mathrm{PRR})=\sqrt{\left(\frac{1}{a}-\frac{1}{a+b}+\frac{1}{c}-\frac{1}{c+d}\right)}$	a ⩾ 3 and 95% CI > 1
	$95\%\mathrm{CI}=e^{\ln (\mathrm{PRR})\pm 1.96\sqrt{\left(\frac{1}{a}-\frac{1}{a+b}+\frac{1}{c}-\frac{1}{c+d}\right)}}$
	$\mathrm{IC}={\log }_2\frac{a(a+b+c+d)}{(a+b)(a+c)}$
BCPNN	$E(\mathrm{IC})={\log }_2\frac{(a+\gamma 11)(a+b+c+d+\alpha )(a+b+c+d+\beta )}{(a+b+c+d+\gamma )(a+b+\alpha 1)(a+c+\beta 1)}$	(1) No signal (−): E(IC) ⩽ 0
	$V(\mathrm{IC})=\frac{1}{{(\ln 2)}^2}\left\{\begin{array}{l}\left[\frac{(a+b+c+d)-a+\gamma -\gamma 11}{(a+\gamma 11)(1+a+b+c+d+\gamma )}\right]\\ +\left[\frac{(a+b+c+d)-(a+b)+a-\alpha 1}{(a+b+\alpha 1)(1+a+b+c+d+\alpha )}\right]\\ +\left[\frac{(a+b+c+d)-(a+c)+\beta -\beta 1}{(a+c+\beta 1)(1+a+b+c+d+\beta )}\right]\end{array}\right\}$	(2) Low signal (+): 0 < E(IC) ⩽ 1.5
	$\gamma =\gamma 11\frac{(a+b+c+d+\alpha )(a+b+c+d+\beta )}{(a+b+\alpha 1)(a+c+\beta 1)}$	(3) Medium signal (++): 1.5 < E(IC) ⩽ 3
	$\mathrm{IC}-2\mathrm{SD}=E(\mathrm{IC})-2\sqrt{V(\mathrm{IC})}$	(4) High signal (+++): E(IC) > 3
	$\gamma 11=1$
	Where α1 = β1 = 1; α = β = 2;
MGPS	$\mathrm{EBGM}=\frac{a(a+b+c+d)}{(a+c)(a+b)}$	EBGM05 > 2 and a > 0
	$\mathrm{EBGM}05=e^{\ln (\mathrm{EBGM})-\sqrt[2]{1.64\left(\frac{1}{a}+\frac{1}{b}+\frac{1}{c}+\frac{1}{d}\right)}}$

BCPNN, Bayesian confidence propagation neural network; CI, confidence interval; EBGM, empirical Bayesian geometric mean; IC, information component; MGPS, multi-item gamma Poisson shrinker; PRR, proportional reported ratio; ROR, reporting odds ratio.