Table 1 Summary of the main and external datasets.

From: Artificial intelligence-assisted prediction of Demodex mite density in facial erythema

Characteristics

Dataset

Main

External

Data collection period

2016. 1–2022. 12

2020. 3–2023. 8

Location (hospital)

Department of dermatology, Severance Hospital

Department of dermatology, Yongin Severance Hospital

Dataset allocation

Training (80%)

External testing (100%)

Validation (10%)

 

(Internal) Testing (10%)

 

Camera type

Digital camera (Canon EOS RP 24–105 mm; 26.2 megapixels)

Digital camera (Canon EOS 800D, 24.2 megapixels)

Lighting condition

Standardized indoor clinical lighting with a uniform blue background and white overhead illumination

Patient demographics

  

 Unique individuals, n

1024

100

 Female sex

697 (68.1)

70 (70.0)

 Age at diagnosis

32.5 (24.0–47.0)

37.5 (28.0–53.0)

High Demodex density

255 (24.9)

45 (45.0)

Associated symptomsa

  

 Flushing

434 (42.4)

49 (49.0)

 Itching

828 (80.9)

78 (78.0)

 Burning/stinging

223 (21.8)

30 (30.0)

 Edema

127 (12.4)

21 (21.0)

 Dry sense

154 (15.0)

19 (19.0)

Positive patch test

316 (30.9)

41 (41.0)

Extrafacial skin involvement

490 (47.9)

26 (26.0)

Serum allergy marker

  

 ECP (µg/L)

26.2 (17.0–41.0)

22.7 (16.2–31.7)

 Eosinophil count (cells/µL)

160.0 (80.0–310.0.0.0)

111.2 (67.5–205.0)

 Total IgE (IU/mL)

111.2 (35.0–461.9.0.9)

66.7 (27.7–205.0)

  1. ECP, eosinophil cationic protein; IgE, immunoglobulin E. Data are presented as n (%) or median (interquartile range)aPatients might be listed in >1 category