A computational clinical decision-supporting system to suggest effective anti-epileptic drugs for pediatric epilepsy patients based on deep learning models using patient’s medical history

Cho, Daeahn; Yu, Myeong-Sang; Shin, Jeongyoon; Lee, Jingyu; Kim, Yubin; Kang, Hoon-Chul; Kim, Se Hee; Na, Dokyun

doi:10.1186/s12911-024-02552-w

BMC Medical Informatics and Decision Making

Table 3 The results of 10-fold cross-validation of our models and conventional machine learning models for each AED.

From: A computational clinical decision-supporting system to suggest effective anti-epileptic drugs for pediatric epilepsy patients based on deep learning models using patient’s medical history

AEDs	Models	AUROC	BAL-ACC	SEN	SPE	PPV	NPV
Vigabatrin	Our model	0.90 ± 0.003	0.93 ± 0.030	0.88 ± 0.057	0.99 ± 0.003	0.94 ± 0.037	1.00 ± 0.000
	KNN	0.79 ± 0.024	0.88 ± 0.025	0.78 ± 0.047	0.98 ± 0.003	0.58 ± 0.047	1.00 ± 0.000
	Logistic regression	0.77 ± 0.015	0.60 ± 0.006	0.22 ± 0.009	0.99 ± 0.003	0.66 ± 0.037	0.89 ± 0.006
	Naïve Bayes	0.74 ± 0.022	0.54 ± 0.003	0.10 ± 0.003	0.99 ± 0.003	0.77 ± 0.047	0.72 ± 0.015
	Random forest	0.81 ± 0.028	0.83 ± 0.052	0.68 ± 0.101	0.99 ± 0.003	0.65 ± 0.056	0.98 ± 0.009
	LightGBM	0.82 ± 0.028	0.92 ± 0.029	0.85 ± 0.056	0.99 ± 0.003	0.65 ± 0.056	0.99 ± 0.003
Prednisolone	Our model	0.80 ± 0.047	0.92 ± 0.031	0.85 ± 0.063	1.00 ± 0.000	0.91 ± 0.126	1.00 ± 0.000
	KNN	0.67 ± 0.072	0.83 ± 0.036	0.68 ± 0.069	0.99 ± 0.003	0.50 ± 0.142	1.00 ± 0.000
	Logistic regression	0.63 ± 0.069	0.65 ± 0.052	0.31 ± 0.101	0.99 ± 0.003	0.35 ± 0.142	0.92 ± 0.012
	Naïve Bayes	0.62 ± 0.072	0.52 ± 0.004	0.06 ± 0.012	0.99 ± 0.003	0.40 ± 0.126	0.78 ± 0.018
	Random forest	0.70 ± 0.069	0.53 ± 0.015	0.08 ± 0.028	0.99 ± 0.003	0.55 ± 0.148	0.85 ± 0.037
	LightGBM	0.61 ± 0.063	0.52 ± 0.015	0.07 ± 0.027	0.98 ± 0.003	0.30 ± 0.126	0.91 ± 0.012
Clobazam	Our model	0.92 ± 0.050	0.91 ± 0.045	0.82 ± 0.088	1.00 ± 0.003	0.90 ± 0.101	1.00 ± 0.000
	KNN	0.67 ± 0.069	0.82 ± 0.064	0.65 ± 0.126	0.99 ± 0.003	0.35 ± 0.142	1.00 ± 0.00
	Logistic regression	0.64 ± 0.069	0.68 ± 0.018	0.38 ± 0.034	0.99 ± 0.003	0.35 ± 0.142	0.98 ± 0.006
	Naïve Bayes	0.57 ± 0.066	0.50 ± 0.004	0.02 ± 0.006	0.99 ± 0.003	0.45 ± 0.148	0.69 ± 0.006
	Random forest	0.70 ± 0.072	0.56 ± 0.031	0.14 ± 0.060	0.99 ± 0.003	0.50 ± 0.142	0.91 ± 0.015
	LightGBM	0.69 ± 0.069	0.54 ± 0.040	0.09 ± 0.037	0.99 ± 0.003	0.50 ± 0.142	0.88 ± 0.027

* Mean ± standard error

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com