Skip to main content
Fig. 2 | BMC Medical Research Methodology

Fig. 2

From: Unsupervised anomaly detection of implausible electronic health records: a real-world evaluation in cancer registries

Fig. 2

Precision \(\frac{\#impl}{n}\) over the number n of selected highest-ranked (most anomalous) records in the entire dataset. We selected the n most anomalous records of the entire dataset according to the anomaly score of each method. For less than 100 selected records, FindFPOF finds more implausible records than the autoencoder. For more than 100 selected records, the performance of FindFPOF and the autoencoder is approximately equal

Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us