Predicting patient-reported outcome of activities of daily living in stroke rehabilitation: a machine learning study

Chen, Yu-Wen; Lin, Keh-chung; Li, Yi-chun; Lin, Chia-Jung

doi:10.1186/s12984-023-01151-6

Journal of NeuroEngineering and Rehabilitation

Table 3 Model performance metrics and training and validation scores for the predictive models

From: Predicting patient-reported outcome of activities of daily living in stroke rehabilitation: a machine learning study

Model	Accuracy	AUC	Specificity	Sensitivity	NPV	PPV	Train score	Validation, median (IQR)
MAL-AOU
18 features
LR	0.72	0.74	0.56	0.81	0.63	0.76	0.66	0.60 (0.19)
KNN	0.56	0.48	0.33	0.69	0.38	0.65	0.75	0.60 (0.17)
SVM	0.60	0.56	0.33	0.75	0.43	0.67	0.86	0.63 (0.10)
RF	0.68	0.76	0.44	0.88	0.67	0.74	1.00	0.65 (0.18)
6 features
LR	0.68	0.74	0.78	0.63	0.54	0.83	0.59	0.47 (0.28)
KNN	0.52	0.66	0.44	0.56	0.36	0.64	1.00	0.65 (0.18)
SVM	0.60	0.69	0.56	0.63	0.45	0.71	0.89	0.65 (0.25)
RF	0.72	0.80	0.67	0.75	0.60	0.80	1.00	0.70 (0.20)
5 features
LR	0.60	0.69	0.78	0.50	0.47	0.80	0.57	0.58 (0.25)
KNN	0.60	0.56	0.44	0.69	0.44	0.69	1.00	0.60 (0.16)
SVM	0.52	0.77	0.78	0.38	0.41	0.75	0.57	0.50 (0.18)
RF	0.64	0.69	0.33	0.81	0.50	0.68	1.00	0.70 (0.09)
4 features
LR	0.60	0.70	0.67	0.56	0.46	0.75	0.58	0.53 (0.28)
KNN	0.56	0.60	0.33	0.69	0.38	0.65	1.00	0.65 (0.19)
SVM	0.64	0.74	0.67	0.63	0.50	0.77	0.61	0.60 (0.15)
RF	0.64	0.64	0.44	0.75	0.50	0.71	1.00	0.74 (0.18)
MAL-QOM
18 features
LR	0.76	0.81	0.67	0.81	0.67	0.81	0.75	0.68 (0.10)
KNN	0.72	0.78	0.56	0.81	0.63	0.76	0.78	0.50 (0.19)
SVM	0.76	0.83	0.78	0.75	0.64	0.86	0.77	0.68 (0.10)
RF	0.76	0.83	0.67	0.81	0.67	0.81	1.00	0.50 (0.19)
6 features
LR	0.60	0.71	0.67	0.56	0.46	0.75	0.63	0.60 (0.26)
KNN	0.52	0.57	0.56	0.50	0.38	0.67	0.73	0.65 (0.10)
SVM	0.64	0.49	0.33	0.81	0.50	0.68	0.97	0.60 (0.10)
RF	0.52	0.67	0.67	0.50	0.43	0.73	0.81	0.60 (0.19)
5 features
LR	0.60	0.72	0.67	0.56	0.46	0.75	0.61	0.60 (0.15)
KNN	0.76	0.75	0.56	0.88	0.71	0.78	1.00	0.60 (0.26)
SVM	0.52	0.62	0.67	0.44	0.40	0.70	0.81	0.60 (0.26)
RF	0.60	0.71	0.56	0.63	0.45	0.71	0.85	0.70 (0.16)
4 features
LR	0.56	0.72	0.56	0.56	0.42	0.69	0.67	0.60 (0.19)
KNN	0.60	0.62	0.33	0.75	0.43	0.67	1.00	0.70 (0.23)
SVM	0.60	0.71	0.67	0.50	0.43	0.73	0.77	0.70 (0.10)
RF	0.72	0.75	0.67	0.75	0.60	0.80	0.99	0.70 (0.21)
NEADL
18 features
LR	0.56	0.57	0.69	0.33	0.65	0.38	0.62	0.50 (0.26)
KNN	0.52	0.41	0.63	0.33	0.63	0.33	0.97	0.60 (0.25)
SVM	0.60	0.65	0.94	0.00	0.63	0.00	0.67	0.65 (0.10)
RF	0.76	0.81	0.75	0.78	0.86	0.64	0.81	0.70 (0.16)
6 features
LR	0.52	0.57	0.63	0.33	0.63	0.33	0.60	0.65 (0.20)
KNN	0.56	0.48	0.63	0.44	0.67	0.40	0.95	0.60 (0.20)
SVM	0.64	0.62	0.50	0.89	0.89	0.50	0.55	0.70 (0.18)
RF	0.72	0.85	0.75	0.67	0.80	0.60	0.80	0.70 (0.16)
5 features
LR	0.64	0.72	0.75	0.44	0.71	0.50	0.62	0.65 (0.20)
KNN	0.64	0.63	0.69	0.56	0.73	0.50	0.94	0.60 (0.10)
SVM	0.64	0.76	1.00	0.00	0.64	0.00	0.70	0.60 (0.09)
RF	0.68	0.82	0.75	0.56	0.75	0.56	0.86	0.68 (0.18)
4 features
LR	0.64	0.72	0.75	0.44	0.71	0.50	0.65	0.60 (0.20)
KNN	0.68	0.71	0.75	0.56	0.75	0.56	0.93	0.68 (0.20)
SVM	0.60	0.70	0.63	0.56	0.71	0.45	0.62	0.60 (0.28)
RF	0.76	0.87	0.75	0.78	0.86	0.64	0.80	0.70 (0.18)
SIS-ADL
18 features
LR	0.92	0.98	0.94	0.86	0.94	0.86	0.98	0.90 (0.08)
KNN	0.80	0.75	0.94	0.43	0.81	0.75	0.96	0.68 (0.10)
SVM	0.96	0.96	1.00	0.86	0.95	1.00	0.95	0.90 (0.15)
RF	0.68	0.76	0.83	0.29	0.75	0.40	1.00	0.70 (0.09)
6 features
LR	0.72	0.80	0.83	0.43	0.79	0.50	0.77	0.75 (0.27)
KNN	0.72	0.77	0.72	0.71	0.87	0.50	0.78	0.70 (0.06)
SVM	0.76	0.82	0.83	0.57	0.83	0.57	0.77	0.70 (0.18)
RF	0.68	0.72	0.72	0.57	0.81	0.44	0.87	0.65 (0.19)
5 features
LR	0.80	0.81	0.89	0.57	0.84	0.67	0.75	0.65 (0.19)
KNN	0.76	0.76	0.83	0.57	0.83	0.57	0.73	0.65 (0.10)
SVM	0.84	0.92	0.83	0.86	0.94	0.67	0.72	0.70 (0.13)
RF	0.68	0.74	0.78	0.43	0.78	0.43	0.80	0.65 (0.16)
4 features
LR	0.76	0.87	0.83	0.57	0.83	0.57	0.75	0.70 (0.09)
KNN	0.68	0.69	0.78	0.43	0.78	0.43	0.82	0.70 (0.08)
SVM	0.72	0.88	0.67	0.86	0.92	0.50	0.72	0.74 (0.20)
RF	0.64	0.72	0.72	0.43	0.76	0.38	0.90	0.70 (0.19)

IQR interquartile range, MAL Motor Activity Log, AOU Amount of Use, QOM Quality of Movement, NEADL Nottingham Extended Activities of Daily Living, SIS-ADL Stroke Impact Scale Activities of Daily Living domain, LR logistic regression, KNN k-nearest neighbors, SVM support vector machine, RF random forest, AUC area under the receiver operating characteristic curve, NPV negative predictive value, PPV positive predictive value

Back to article page

ISSN: 1743-0003

Contact us

Submission enquiries: journalsubmissions@springernature.com