Cervical cancer prognosis and diagnosis using electrical impedance spectroscopy

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

AUC values for testing sets from 10 repeated two-fold cross validation runs with three logistic regression models

Repetitions	R¯0,α¯,ΔR0,CI,Ref $\begin{align}& {{{\bar{R}}}_{0}},\bar{\alpha },\Delta {{R}_{0}}, \\ & \,\,\text{CI,Ref} \\ \end{align}$	R¯02,α¯2,ΔR02,CI,Ref $\begin{align}& \bar{R}_{0}^{2},{{{\bar{\alpha }}}^{2}},\Delta R_{0}^{2}, \\ & \,\,\,\,\,\,\,\text{CI,Ref} \\ \end{align}$	R¯03,α¯3,ΔR03,CI, Ref $\begin{align}& \bar{R}_{0}^{3},{{{\bar{\alpha }}}^{3}},\Delta R_{0}^{3}, \\ & \,\,\,\,\,\,\text{CI, Ref} \\ \end{align}$
1	0.9127	0.9160	0.9165
2	0.9177	0.9178	0.9190
3	0.9013	0.9034	0.9045
4	0.9165	0.9210	0.9238
5	0.8830	0.8840	0.8858
6	0.9206	0.9222	0.9230
7	0.9164	0.9165	0.9172
8	0.9053	0.9061	0.9075
9	0.9122	0.9146	0.9155
10	0.9181	0.9215	0.9222
Mean AUC	0.9104	0.9123	0.9135

Mean AUC values from 100 5-fold cross validation runs with linear logistic regression models

Feature combinations	Mean AUC	Feature combinations	Mean AUC
α¯,Δα $\bar{\alpha },\Delta \alpha $	0.5870	R¯0,α,Δα ${{\bar{R}}_{0}},\,\alpha ,\,\Delta \alpha $	0.5723
α¯,ΔR∞ $\bar{\alpha },\Delta {{R}_{\infty }}$	0.5777	f¯c,α,Δα ${{\bar{f}}_{c}},\alpha ,\Delta \alpha $	0.5716
f¯c,α¯ ${{\bar{f}}_{c}},\bar{\alpha }$	0.5745	α¯,ΔR0 $\bar{\alpha },\Delta {{R}_{0}}$	0.5715
fc¯,Δα $\overline{{{f}_{c}}},\,\Delta \alpha $	0.5744	α¯,Δfc $\bar{\alpha },\Delta {{f}_{c}}$	0.5686
f¯c,ΔR∞,Δα ${{\bar{f}}_{c}},\Delta {{R}_{\infty }},\Delta \alpha $	0.5736	R¯0,α¯ $$\bar{R}_{0}, \bar{\alpha}$$	0.5678

p-values from MANOVA using EIS data taken from 1704 women for HG CIN detection

Feature combinations	p-values	Feature combinations	p-values
R¯0,α¯,ΔR0 ${{\bar{R}}_{0}},\bar{\alpha },\Delta {{R}_{0}}$	1.1003 × 10−31	R¯0,α¯,ΔR0,Δfc ${{\bar{R}}_{0}},\bar{\alpha },\Delta {{R}_{0}},\Delta {{f}_{c}}$	5.0124 × 10−31
R¯0,ΔR0 ${{\bar{R}}_{0}},\Delta {{R}_{0}}$	1.5861 × 10−31	R¯∞,R¯0,ΔR∞,ΔR0 ${{\bar{R}}_{\infty }},{{\bar{R}}_{0}},\Delta {{R}_{\infty }},\Delta {{R}_{0}}$	5.3276 × 10−31
R¯0,α¯,ΔR∞,ΔR0 ${{\bar{R}}_{0}},\bar{\alpha },\Delta {{R}_{\infty }},\Delta {{R}_{0}}$	2.6287 × 10−31	R¯0,ΔR0,Δfc ${{\bar{R}}_{0}},\Delta {{R}_{0}},\Delta {{f}_{c}}$	6.5955 × 10−31
R¯0,ΔR∞,ΔR0 ${{\bar{R}}_{0}},\Delta {{R}_{\infty }},\Delta {{R}_{0}}$	3.1665 × 10−31	R¯0,ΔR0,Δα ${{\bar{R}}_{0}},\Delta {{R}_{0}},\Delta \alpha $	7.2683 × 10−31
R¯0,α¯,ΔR0,Δα ${{\bar{R}}_{0}},\bar{\alpha },\Delta {{R}_{0}},\Delta \alpha $	3.4687 × 10−31	R¯0,f¯c,α¯,ΔR0 ${{\bar{R}}_{0}},{{\bar{f}}_{c}},\bar{\alpha },\Delta {{R}_{0}}$	7.4318 × 10−31

Mean AUC values from 100 5-fold cross validation runs with nonlinear logistic regression models

Feature combinations	Mean AUC	Feature combinations	Mean AUC
α¯2,Δα2 ${{\bar{\alpha }}^{2}},\Delta {{\alpha }^{2}}$	0.6103	f¯c,α¯2,Δα2 ${{\bar{f}}_{c}},{{\bar{\alpha }}^{2}},\Delta {{\alpha }^{2}}$	0.5911
Δα,α¯2 $\Delta \alpha ,{{\bar{\alpha }}^{2}}$	0.5992	R¯02,α2,Δα2 $\bar{R}_{0}^{2},{{\alpha }^{2}},\Delta {{\alpha }^{2}}$	0.5899
α¯,Δα2 $\bar{\alpha },\Delta {{\alpha }^{2}}$	0.5989	ΔR∞,α¯2,Δα2 $\Delta {{R}_{\infty }},{{\bar{\alpha }}^{2}},\Delta {{\alpha }^{2}}$	0.5895
α¯2,α¯⋅Δα,Δα2 ${{\bar{\alpha }}^{2}},\bar{\alpha }\cdot \Delta \alpha ,\Delta {{\alpha }^{2}}$	0.5946	ΔR∞2,α¯2,Δα2 $\Delta R_{\infty }^{2},{{\bar{\alpha }}^{2}},\Delta {{\alpha }^{2}}$	0.5891
α,Δα,Δα2 $\alpha ,\,\Delta \alpha ,\,\Delta {{\alpha }^{2}}$	0.5939	α¯2,Δfc,Δα2 ${{\bar{\alpha }}^{2}},\Delta {{f}_{c}},\Delta {{\alpha }^{2}}$	0.5885

Regression coefficient estimates and the associated p-values for the final logistic regression model

β	estimates	p-values
βo	-2.9619	3.9518 × 10−32
β1	−7.4684 × 10−11	0.0047
β2	3.3987	0.0090
β1	3.0025 × 10−11	0.0044
βCI	2.3621	3.9281 × 10−47
βRef	2.2241	5.8068 × 10−35

Classification performance comparison between the new classifier developed and the previous classifiers

Classifier	AUC	Sensitivity	Specificity
Logistic regression (α¯2,Δα2) $\left( {{{\bar{\alpha }}}^{2}},\Delta {{\alpha }^{2}} \right)$	0.628	45.714%	82.022%
Impedance at 152Hz	0.621	38.7%	83.4%
Slope (between 1.22	0.596	45.2%	70.1%
and 2.44kHz) as α

p-values from MANOVA using EIS data taken at initial colposcopy of 569 women for evaluation of prognostic value of EIS

Feature combinations	p-values	Feature combinations	p-values
α¯,Δα $\bar{\alpha },\Delta \alpha $	0.0168	f¯c,Δα ${{\bar{f}}_{c}},\Delta \alpha $	0.0286
α¯,ΔR0 $\bar{\alpha },\Delta {{R}_{0}}$	0.0231	R¯0,α¯ ${{\bar{R}}_{0}},\bar{\alpha }$	0.0295
f¯c,α¯ ${{\bar{f}}_{c}},\bar{\alpha }$	0.0256	R¯∞,α¯ ${{\bar{R}}_{\infty }},\bar{\alpha }$	0.0296
α¯,ΔR∞ $\bar{\alpha },\Delta {{R}_{\infty }}$	0.0274	f¯c,α¯,Δα ${{\bar{f}}_{c}},\bar{\alpha },\Delta \alpha $	0.0314
α¯,Δfc $\bar{\alpha },\Delta {{f}_{c}}$	0.0275	R¯0,α¯,Δα ${{\bar{R}}_{0}},\bar{\alpha },\Delta \alpha $	0.0335