Table 4 Performance of 10 ML algorithms in the training and validation cohorts for early diagnosis of BC.

From: Machine learning model for early diagnosis of breast cancer based on PiRNA expression with CA153

	Name	Optimal threshold	Accuracy	Positive Precision	Negative Precision	Positive Recall	Negative Recall	F1 score	kappa	AUC	AUC 95% CI
train	AdaBoost	0.49845	0.980769	0.972222	0.988095	0.985915493	0.976470588	0.98078	0.961271	0.999171	0.9968582964289002-1.0
train	ANN	0.517937	0.794872	0.867925	0.757282	0.647887324	0.917647059	0.789803	0.577594	0.812593	0.7392675981258746–0.8760783996022822
train	DT	0.5	0.788462	0.88	0.745283	0.61971831	0.929411765	0.781734	0.56284	0.788152	0.7245831383806066–0.8474177767426827
train	GBDT	0.464509	0.99359	1	0.988372	0.985915493	1	0.993586	0.98706	1	0.9999999999999999-1.0
train	KNN	0.5	0.769231	0.857143	0.728972	0.591549296	0.917647059	0.761298	0.52253	0.834963	0.768916320895988–0.8909322273028321
train	LGBM	0.400531	0.839744	0.838235	0.840909	0.802816901	0.870588235	0.839404	0.675757	0.91251	0.863754546957672–0.9525539685923516
train	LR	0.510033	0.75641	0.770492	0.747368	0.661971831	0.835294118	0.75395	0.503101	0.799171	0.7230585237547262–0.8629722618044392
train	RF	0.394845	0.814103	0.783784	0.841463	0.816901408	0.811764706	0.814356	0.626486	0.893538	0.8455313358116229–0.9405541361030084
train	SVM	0.5	0.769231	0.818182	0.742574	0.633802817	0.882352941	0.764504	0.525916	0.806628	0.7299192040598291–0.8709271904608475
train	XGBoost	0.68755	0.99359	1	0.988372	0.985915493	1	0.993586	0.98706	1	0.9999999999999999-1.0
Validation	AdaBoost	0.494679	0.691176	0.625	0.785714	0.806451613	0.594594595	0.68937	0.391823	0.737576	0.6026757097069598–0.8455445075757576
Validation	ANN	0.42376	0.75	0.769231	0.738095	0.64516129	0.837837838	0.746946	0.489399	0.759372	0.6181468229002831–0.8791060102688009
Validation	DT	0.883721	0.544118	0	0.544118	0	1	0.383473	0	0.686138	0.5806740723045071–0.7938135915958496
Validation	GBDT	0.710062	0.764706	0.857143	0.723404	0.580645161	0.918918919	0.756087	0.512981	0.841325	0.7338505747126437–0.9260123026252057
Validation	KNN	0.6	0.691176	0.8125	0.653846	0.419354839	0.918918919	0.667921	0.352087	0.771578	0.6635252157119272–0.8704733896072797
Validation	LGBM	0.386878	0.823529	0.827586	0.820513	0.774193548	0.864864865	0.82291	0.642419	0.839146	0.7289324325847764–0.9324775529614239
Validation	LR	0.519913	0.764706	0.826087	0.733333	0.612903226	0.891891892	0.758754	0.515583	0.804708	0.6856231279418057–0.905282741738066
Validation	RF	0.486437	0.794118	0.84	0.767442	0.677419355	0.891891892	0.790809	0.578388	0.841761	0.7331673385100805–0.9250475778546713
Validation	SVM	0.289992	0.735294	0.666667	0.827586	0.838709677	0.648648649	0.734377	0.47737	0.809939	0.6940746753246754–0.908764714600271
Validation	XGBoost	0.472235	0.794118	0.774194	0.810811	0.774193548	0.810810811	0.794118	0.585004	0.842197	0.7381517033690946–0.9308225108225108

K-nearest Neighbor (KNN), logistic regression (LR), random forest (RF), decision tree (DT), artificial neural networks (ANN), support vector machine (SVM), gradient boosting decision tree (GBDT), light gradient boosting machine (LGBM), adaptive boosting (AdaBoost), extreme gradient boosting (XGboost).

Back to article page

Table 4 Performance of 10 ML algorithms in the training and validation cohorts for early diagnosis of BC.

Search

Quick links