Development of machine learning models aiming at knee osteoarthritis diagnosing: an MRI radiomics analysis

Cui, Tingrun; Liu, Ruilong; Jing, Yang; Fu, Jun; Chen, Jiying

doi:10.1186/s13018-023-03837-y

Journal of Orthopaedic Surgery and Research

Table 3 Results of algorithms of KNN, LR and SVM

From: Development of machine learning models aiming at knee osteoarthritis diagnosing: an MRI radiomics analysis

Algorithm	Model	Cohort	AUC (95% CI)	Accuracy	Sensitivity	Specificity
KNN	M-S	Train	0.786 (0.710–0.844)	0.701	0.614	0.783
		Validation	0.712 (0.529–0.833)	0.613	0.533	0.688
	M-C	Train	0.805 (0.741–0.863)	0.718	0.544	0.883
		Validation	0.771 (0.621–0.886)	0.645	0.533	0.750
	M-S-C	Train	0.866 (0.809–0.915)	0.786	0.667	0.900
		Validation	0.860 (0.724–0.952)	0.839	0.733	0.938
	L-S	Train	0.832 (0.773–0.888)	0.744	0.667	0.817
		Validation	0.706 (0.530–0.83)	0.677	0.533	0.875
	L-C	Train	0.835 (0.770–0.886)	0.735	0.649	0.817
		Validation	0.752 (0.592–0.895)	0.710	0.667	0.750
	L-S-C	Train	0.867 (0.890–0.912)	0.778	0.632	0.917
		Validation	0.796 (0.625–0.931)	0.839	0.733	0.938
	P-S	Train	0.774 (0.707–0.844)	0.667	0.561	0.767
		Validation	0.694 (0.559–0.867)	0.677	0.553	0.813
	P–T	Train	0.834 (0.762–0.889)	0.769	0.702	0.833
		Validation	0.721 (0.598–0.891)	0.677	0.600	0.750
	P-S-T	Train	0.846 (0.786–0.86)	0.769	0.684	0.850
		Validation	0.950 (0.890–0.993)	0.903	0.933	0.875
	Final-M	Train	0.927 (0.878–0.960)	0.880	0.789	0.967
		Validation	0.938 (0.862–0.988)	0.839	0.800	0.875
	Clnc-M	Train	0.695 (0.622–0.762)	0.684	0.632	0.733
		Validation	0.692 (0.531–0.827)	0.642	0.600	0.688
LR	M-S	Train	0.813 (0.736–0.872)	0.718	0.737	0.700
		Validation	0.883 (0.745–0.996)	0.742	0.733	0.750
	M-C	Train	0.774 (0.696–0.840)	0.726	0.684	0.767
		Validation	0.733 (0.567–0.885)	0.710	0.800	0.625
	M-S-C	Train	0.830 (0.759–0.885)	0.744	0.719	0.767
		Validation	0.875 (0.754–0.962)	0.774	0.733	0.813
	L-S	Train	0.876 (0.819–0.924)	0.795	0.789	0.800
		Validation	0.913 (0.804–0.983)	0.806	0.733	0.875
	L-C	Train	0.839 (0.771–0.895)	0.761	0.702	0.817
		Validation	0.842 (0.704–0.950)	0.742	0.800	0.688
	L-S-C	Train	0.917 (0.873–0.952)	0.821	0.789	0.850
		Validation	0.938 (0.857–0.991)	0.871	0.800	0.938
	P-S	Train	0.884 (0.829–0.931)	0.786	0.754	0.817
		Validation	0.883 (0.858–0.992)	0.806	0.867	0.750
	P–T	Train	0.885 (0.832–0.933)	0.821	0.754	0.883
		Validation	0.908 (0.836–0.982)	0.742	0.667	0.813
	P-S-T	Train	0.977 (0.957–0.993)	0.932	0.947	0.917
		Validation	0.921 (0.906–1.000)	0.806	0.867	0.750
	Final-M	Train	0.984 (0.969–0.995)	0.940	0.877	1.000
		Validation	0.983 (0.957–1.000)	0.968	1.000	0.938
	Clnc-M	Train	0.684 (0.599–0.751)	0.684	0.544	0.817
		Validation	0.644 (0.451–9.782)	0.645	0.533	0.451
SVM	M-S	Train	0.829 (0.752–0.885)	0.752	0.737	0.767
		Validation	0.708 (0.521–0.850)	0.645	0.600	0.689
	M-C	Train	0.883 (0.826–0.929)	0.821	0.772	0.867
		Validation	0.792 (0.647–0.919)	0.742	0.800	0.689
	M-S-C	Train	0.885 (0.822–0.931)	0.769	0.737	0.800
		Validation	0.817 (0.649–0.929)	0.710	0.733	0.688
	L-S	Train	0.920 (0.881–0.953)	0.821	0.789	0.850
		Validation	0.838 (0.693–0.940)	0.774	0.667	0.875
	L-C	Train	0.888 (0.833–0.930)	0.786	0.719	0.850
		Validation	0.829 (0.675–0.947)	0.806	0.800	0.813
	L-S-C	Train	0.941 (0.905–0.970)	0.821	0.754	0.883
		Validation	0.896 (0.765–1.000)	0.839	0.800	0.875
	P-S	Train	0.923 (0.883–0.959)	0.821	0.772	0.867
		Validation	0.867 (0.832–0.986)	0.806	0.733	0.875
	P–T	Train	0.915 (0.864–0.953)	0.838	0.789	0.883
		Validation	0.858 (0.744–0.975)	0.806	0.733	0.875
	P-S-T	Train	0.956 (0.927–0.978)	0.846	0.789	0.900
		Validation	0.879 (0.827–1.000)	0.806	0.733	0.875
	Final-M	Train	0.984 (0.968–0.996)	0.940	0.877	1.000
		Validation	0.958 (0.895–1.000)	0.935	0.933	0.938
	Clnc-M	Train	0.747 (0.674–0.815)	0.667	0.667	0.667
		Validation	0.715 (0.548–0.860)	0.710	0.733	0.688

Back to article page

ISSN: 1749-799X

Contact us

Submission enquiries: journalsubmissions@springernature.com