Comparison between machine learning methods for mortality prediction for sepsis patients with different social determinants

Wang, Hanyin; Li, Yikuan; Naidech, Andrew; Luo, Yuan

doi:10.1186/s12911-022-01871-0

BMC Medical Informatics and Decision Making

Table 5 Pairwise comparisons among different racial groups

From: Comparison between machine learning methods for mortality prediction for sepsis patients with different social determinants

	Asian v.s. Black or African American		Asian v.s. Hispanic or Latino		Asian v.s. Other		Asian v.s. White		Black or African American v.s. Hispanic or Latino
	Observed difference	p_val	Observed difference	p_val	Observed difference	p_val	Observed difference	p_val	Observed difference	p_val
Ridge classifier	0.2572	0.074	0.0605	0.738	0.2824	*0.033*	0.2988	*0.018*	− 0.1967	0.189
Perceptron	0.2827	*0.042*	0.0295	0.883	0.2722	0.051	0.2906	*0.021*	− 0.2531	0.081
Passive-aggressive	0.3114	*0.045*	0.1439	0.432	0.3348	*0.018*	0.3299	*0.008*	− 0.1674	0.238
kNN	0.0686	0.647	− 0.0552	0.763	0.1170	0.380	0.1528	0.224	− 0.1238	0.413
Random forest	− 0.0104	0.916	− 0.0392	0.715	0.1370	0.211	0.0890	0.372	− 0.0287	0.781
LinearSVC_L1	0.2647	0.075	0.0571	0.756	0.2822	*0.043*	0.2991	*0.020*	− 0.2076	0.156
LinearSVC_L2	0.2594	0.084	0.0605	0.752	0.2820	*0.042*	0.2990	*0.019*	− 0.1990	0.179
SGDClassifier_L1	0.2832	0.052	0.0714	0.668	0.2908	*0.036*	0.3057	*0.022*	− 0.2118	0.136
SGDClassifier_L2	0.2813	*0.050*	0.0718	0.692	0.2944	*0.019*	0.3091	*0.015*	− 0.2095	0.151
SGDClassifier_EN	0.2858	0.058	0.0718	0.706	0.2954	*0.035*	0.3086	*0.015*	− 0.2140	0.142
MultinomialNB	0.3468	*0.013*	0.0424	0.800	0.2848	*0.035*	0.2848	*0.029*	− 0.3044	*0.032*
BernoulliNB	0.1940	*0.043*	0.2450	0.068	0.2147	*0.015*	0.1986	*0.021*	0.0510	0.609
Logistic regression	0.2617	0.082	0.0814	0.620	0.2918	*0.037*	0.3049	*0.019*	− 0.1804	0.198
SVC_rbf	0.3127	*0.025*	0.0774	0.653	0.2909	*0.030*	0.3259	*0.013*	− 0.2352	0.093
SVC_poly	0.3005	*0.024*	0.0227	0.889	0.3066	*0.025*	0.3132	*0.016*	− 0.2778	0.056
SVC_sigmoid	0.0402	0.780	0.0736	0.666	0.1244	0.375	0.1551	0.235	0.0334	0.796

	Black or African American v.s. Other		Black or African American v.s. White		Hispanic or Latino v.s. Other		Hispanic or Latino v.s. White		Other v.s. White
	Observed difference	p_val	Observed difference	p_val	Observed difference	p_val	Observed difference	p_val	Observed difference	p_val
Ridge classifier	0.0252	0.781	0.0416	0.557	0.2219	0.103	0.2383	0.055	0.0164	0.783
Perceptron	− 0.0104	0.930	0.0080	0.926	0.2427	0.076	0.2611	*0.032*	0.0184	0.747
Passive-aggressive	0.0234	0.764	0.0186	0.791	0.1908	0.167	0.1860	0.134	− 0.0048	0.931
kNN	0.0484	0.564	0.0841	0.225	0.1721	0.189	0.2079	0.081	0.0358	0.537
Random forest	0.1474	*0.029*	0.0994	0.089	0.1762	0.101	0.1282	0.182	− 0.0480	0.278
LinearSVC_L1	0.0175	0.832	0.0344	0.629	0.2251	0.106	0.2420	0.065	0.0170	0.764
LinearSVC_L2	0.0226	0.792	0.0396	0.585	0.2216	0.088	0.2386	0.065	0.0170	0.756
SGDClassifier_L1	0.0076	0.931	0.0225	0.753	0.2194	0.108	0.2343	0.075	0.0149	0.794
SGDClassifier_L2	0.0131	0.882	0.0278	0.699	0.2226	0.080	0.2373	0.059	0.0147	0.786
SGDClassifier_EN	0.0096	0.932	0.0228	0.765	0.2236	0.088	0.2368	0.070	0.0132	0.830
MultinomialNB	− 0.0620	0.491	− 0.0620	0.425	0.2423	0.073	0.2424	0.053	0.0001	1.000
BernoulliNB	0.0207	0.702	0.0046	0.935	− 0.0303	0.764	− 0.0464	0.607	− 0.0161	0.650
Logistic regression	0.0301	0.705	0.0432	0.579	0.2104	0.130	0.2235	0.083	0.0131	0.827
SVC_rbf	− 0.0218	0.799	0.0133	0.860	0.2135	0.110	0.2485	*0.047*	0.0350	0.527
SVC_poly	0.0060	0.930	0.0127	0.848	0.2838	*0.027*	0.2905	*0.019*	0.0066	0.904
SVC_sigmoid	0.0841	0.286	0.1149	0.110	0.0507	0.727	0.0814	0.544	0.0307	0.584

Observe difference: observed difference in AUC when comparing the performance between the sub-populations; p_val: p-value, p values less than or equal to 0.05 were highlighted; Passive-aggressive: passive-aggressive classifier; kNN: k-Nearest Neighbors; LinearSVC_L1 or _L2: support vector machine with linear kernel coupled with L1 or L2 regularization; SGDClassifier_L1 or _L2 or _EN: stochastic gradient descent with L1 or L2 or Elastic Net regularization; MultinomialNB: Multinomial naïve Bayes; BernoulliNB: Bernoulli naïve Bayes; SVC_rbf or _poly or _sigmoid: support vector machine with rbf kernel or polynomial kernel or sigmoid kernel

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com