Table 20 Population-level screening performance under natural prevalence conditions.

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

Population context

Sample size

True positive rate (95% CI)

False positive rate (95% CI)

Positive predictive value (95% CI)

Negative predictive value (95% CI)

Number needed to screen (95% CI)

Cost per detection (USD, 95% CI)

Likelihood ratio+

p-value

Rural community (high prevalence)

15,000

87.3% (85.1–89.5)

2.8% (2.1–3.5)

73.4% (70.8–76.0)

97.8% (97.2–98.4)

23 (19–28)

$127 ($108-$146)

31.2

< 0.001

Urban community (moderate prevalence)

22,000

84.7% (82.9–86.5)

3.2% (2.6–3.8)

68.9% (66.7–71.1)

96.4% (95.9–96.9)

31 (27–36)

$156 ($139-$173)

26.5

< 0.001

Mobile screening (variable prevalence)

8500

78.9% (76.4–81.4)

4.1% (3.3–4.9)

58.7% (55.8–61.6)

94.9% (94.1–95.7)

47 (39–56)

$203 ($176-$230)

19.2

< 0.001

School-based screening (low prevalence)

12,000

71.2% (68.1–74.3)

2.6% (2.0-3.2)

45.3% (41.8–48.8)

98.7% (98.3–99.1)

89 (72–108)

$298 ($259-$337)

27.4

< 0.001

Elderly population (high prevalence)

9200

89.1% (87.2–91.0)

3.7% (2.9–4.5)

76.8% (74.1–79.5)

96.2% (95.5–96.9)

18 (15–22)

$109 ($92-$126)

24.1

< 0.001

International sites (mixed prevalence)

18,400

82.6% (80.9–84.3)

3.4% (2.8-4.0)

64.2% (61.9–66.5)

95.8% (95.3–96.3)

38 (33–44)

$174 ($155-$193)

24.3

< 0.001