Table 27 Independent dataset characteristics and natural variability assessment.

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

Dataset

Source type

Collection period

Natural degradation factors

Equipment age range (mean ± SD)

Operator variability (n)

Patient demographics

Original purpose

Sample quality score (mean ± SD)

Institutional review board status

ExtVal-1 (rural)

Ongoing screening program

2019–2023

Equipment aging, positioning errors

8.7 ± 3.2 years (3–15 years)

12 different operators

Rural farming communities (n = 8340)

Community health screening

3.2 ± 0.8/5.0

Approved (IRB-2019-045)

ExtVal-2 (urban)

Clinical practice archive

2020–2024

High throughput, variable protocols

6.3 ± 2.9 years (2–12 years)

18 different operators

Low-income urban population (n = 12,480)

Routine clinical diagnosis

3.6 ± 0.7/5.0

Approved (IRB-2020-128)

ExtVal-3 (mobile)

Mobile screening consortium

2021–2024

Portable equipment limitations

4.1 ± 2.3 years (1–8 years)

25 different operators

Underserved communities (n = 6580)

Public health outreach

2.7 ± 0.9/5.0

Approved (IRB-2021-074)

ExtVal-4 (international)

Multi-site research network

2018–2023

Cross-population variations

9.2 ± 3.8 years (4–16 years)

34 different operators

Six countries, diverse ethnicities (n = 9520)

Epidemiological research

3.4 ± 0.6/5.0

Multi-site approval

ExtVal-5 (pathology)

Specialized clinical centers

2017–2024

Early-stage detection challenges

8.1 ± 2.7 years (5–14 years)

Expert radiologists (8)

High-risk patients (n = 4890)

Specialized diagnosis

4.1 ± 0.4/5.0

Approved (IRB-2017-156)

ExtVal-6 (emergency)

Emergency department archive

2020–2024

Suboptimal acquisition conditions

7.4 ± 4.2 years (2–18 years)

Non-dental personnel (22)

Emergency presentations (n = 7440)

Urgent care diagnosis

2.9 ± 1.0/5.0

Approved (IRB-2020-203)

ExtVal-7 (longitudinal)

Long-term follow-up study

2018–2024

Temporal equipment changes

8.8 ± 2.1 years (6–12 years)

15 different operators

Tracked patient cohort (n = 5970)

Disease progression monitoring

3.7 ± 0.5/5.0

Approved (IRB-2018-089)