Table 22 Expert clinical assessment of synthetic data quality.

From: Novel metaheuristic optimized latent diffusion framework for automated oral disease detection in public health screening

Assessment criterion

Excellent (%)

Good (%)

Acceptable (%)

Poor (%)

Unacceptable (%)

Clinical utility score (mean ± SD)

Inter-rater reliability (κ)

p-value

Pathological authenticity

67.3

24.8

6.2

1.4

0.3

4.57 ± 0.42/5.0

0.847

< 0.001

Anatomical consistency

72.1

21.4

5.1

1.2

0.2

4.63 ± 0.38/5.0

0.891

< 0.001

Diagnostic relevance

64.9

26.7

6.8

1.4

0.2

4.54 ± 0.45/5.0

0.823

< 0.001

Morphological diversity

58.3

29.4

9.7

2.3

0.3

4.43 ± 0.51/5.0

0.796

< 0.001

Structural variation

61.7

27.8

8.1

2.1

0.3

4.48 ± 0.48/5.0

0.812

< 0.001

Feature complexity

59.4

28.9

9.2

2.2

0.3

4.45 ± 0.49/5.0

0.805

< 0.001

Overall clinical value

63.2

26.1

8.4

2.0

0.3

4.51 ± 0.46/5.0

0.834

< 0.001

Back to article page