Table 5. Multiple Regression Models and Performance in Predicting Gestational Age and Preterm Birth.

Regression model; total number	% (n)	MSE	Sensitivity	Specificity	PPV	NPV	Accuracy	F
(a) LMP and SFH (linear); n = 672	3.1	75.4	0.47	0.99	0.76	0.97	0.97	0.58
	(21)		(±0.17)	(±0.007)	(±0.18)	(±0.01)	(±0.01)
(b) LMP and total BS; n = 608	0.5	99.8	0.10	1	1	0.96	0.96	0.18
	(3)		(±0.10)	(±0)	(±0)	(±0.02)	(±0.02)
(c) SFH (linear) and total BS; n = 623	2.1	69.3	0.35	0.99	0.84	0.97	0.96	0.50
	(13)		(±0.2)	(±0.004)	(±0.20)	(±0.01)	(±0.01)
(d) LMP, SFH (linear) and total BS; n = 608	2.3	69.1	0.40	0.99	0.86	0.97	0.97	0.54
	(14)		(±0.18)	(±0.004)	(±0.18)	(±0.01)	(±0.01)
(e) LMP and SFH (sequential); n = 491	1.63	77.5	0.22	0.99	0.75	0.96	0.95	0.34
	(8)		(±0.16)	(±0.006)	(±0.30)	(±0.02)	(±0.02)
(f) LMP* and total BS; n = 301	0.7	72.5	0.15	1	1	0.96	0.96	0.27
	(2)		(±0.20)	(±0)	(±0)	(±0.02)	(±0.02)

Note: GW, gestational weeks; LMP, last menstrual period; LMP*, corrected last menstrual period based on PNG guidelines; SFH, symphysis-pubis fundal height; BS, Ballard Score PPV, positive predictive value; NPV, negative predictive value; F, F-measure. Numbers in parentheses under sensitivity, specificity, PPV, NPV and accuracy are 95% confidence intervals. The mean square error (MSE) assesses the fit of the regression models used to predict gestational age while sensitivity, specificity, PPV, NPV, accuracy and F-measure are measures of performance in predicting PTB.