Table 18 Comparison of artificial training balance vs. natural disease prevalence.

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

Pathology type

Training distribution (%)

Natural prevalence (mean ± SD)

Prevalence range (95% CI)

Overrepresentation factor

χ² statistic

p-value

Dental caries (early)

8.0

35.7 ± 15.2%

15.2–67.3%

0.22x

127.45

< 0.001

Dental caries (advanced)

8.0

23.4 ± 9.8%

8.9–41.7%

0.34x

89.32

< 0.001

Periodontal disease

8.3

28.9 ± 10.6%

12.6–48.3%

0.29x

98.76

< 0.001

Apical periodontitis

8.3

12.1 ± 5.4%

4.7–22.8%

0.69x

12.34

< 0.001

Root resorption

8.3

2.3 ± 1.2%

0.8–4.9%

3.61x

145.67

< 0.001

Dental fractures

8.3

8.7 ± 3.6%

3.2–15.4%

0.95x

0.84

0.359

Impacted teeth

8.3

14.2 ± 5.2%

6.8–24.1%

0.58x

23.45

< 0.001

Bone pathology

8.3

3.8 ± 2.0%

1.2–7.9%

2.18x

67.89

< 0.001

Cysts/tumors

8.3

0.9 ± 0.7%

0.1–2.4%

9.22x

234.56

< 0.001

Orthodontic issues

8.3

19.6 ± 6.8%

9.1–32.7%

0.42x

45.67

< 0.001

TMJ disorders

8.3

7.4 ± 3.3%

2.8–13.9%

1.12x

2.15

0.143

Developmental anomalies

8.3

3.1 ± 1.8%

0.9–6.8%

2.68x

78.23

< 0.001