Fig. 2: Classification performance of the proposed semi-supervised method (SSL) vs benchmarking methods. | npj Digital Medicine

Fig. 2: Classification performance of the proposed semi-supervised method (SSL) vs benchmarking methods.

From: Label efficient phenotyping for Long COVID using electronic health records

Fig. 2: Classification performance of the proposed semi-supervised method (SSL) vs benchmarking methods.

Benchmark methods for comparison were U09.9 counts greater or equal to 1, 2, 3 and 4, and unsupervised XGBoost (XGB). For all methods shown, F-Score, TPR, PPV, NPV and prevalence of cases were identified through evaluation against the gold-standard chart review labels using WHO-1 definition at (a) VHA and (b) UPMC. VHA Health Administration, UPMC University of Pittsburgh Medical Center, XGB XGBoost tree models, TPR True Positive Rate, PPV Positive Predictive Value, NPV Negative Predictive Value.

Back to article page