Deep learning computer-aided detection system for pneumonia in febrile neutropenia patients: a diagnostic cohort study

Table 1 Demographic and clinical information

Variable	Data
Male patient (%)^a	52.3% (216/413)
Age (year)^b	59 (48–67)
Number of CXR per patient^a
1 CXR per patient	79.9% (330/413)
2 CXRs per patient	13.6% (56/413)
3 CXRs per patient	6.1% (25/413)
4 CXRs per patient	0.5% (2/413)
Underlying disease of medical history^a
Hematologic malignancy or lymphoma	60.8% (251/413)
Solid organ malignancy	26.4% (109/413)
Hematopoietic stem cell transplantation	17.4% (72/413)
Solid organ transplantation	1.7% (7/413)
ANC (count per mm³)^b	36 (0–166)
Peak body temperature (°C)^b	38.9 (38.5–39.3)
Site of infection^c
Pneumonia	24.4% (128/525)
Gastrointestinal tract	6.9% (36/525)
Bone or soft tissue	4.2% (22/525)
Oral cavity	1.9% (10/525)
Bloodstream infection	1.7% (9/525)
Urinary tract	1.3% (7/525)
Biliary tree	0.8% (4/525)
Unidentifiable infection focus	62.1% (326/525)
CXRs from fixed radiography unit	27.8% (146/525)
CXRs from portable radiography scanner	72.2% (379/525)

CAD, computer-aided detection; CXR, chest X-ray; ANC, absolute neutrophil count
^aData indicate proportions among entire patients (numerators/denominators)
^bData indicate median (interquartile ranges) values
^cData indicate proportions among entire episodes of febrile neutropenia (numerators/denominators)

ISSN: 1471-2466