Skip to main content

Table 2 Simulated indications dataset results

From: Clinical language search algorithm from free-text: facilitating appropriate imaging

Analysis metric

Simple indications

Complex indications

P value

Proportion of ground truth documents in top 3

0.985

0.849

P < 0.0001

Average ground truth rank

1.36

2.66

P < 0.00001

Average NDCG

0.841

0.801

P < 0.0001

  1. Chi-squared test was used to calculate significance of the proportion of ground truth documents in top 3, and Mann–Whitney U test was used for ground truth rank and NDCG. All metrics show that the algorithm performed significantly better on simple indications than on complex ones. Normalized discounted cumulative gain (NDCG), Appropriateness criteria (AC)
Back to article page

BMC Medical Imaging

ISSN: 1471-2342

Contact us