J Clin Med Res

Journal of Clinical Medicine Research, ISSN 1918-3003 print, 1918-3011 online, Open Access

Article copyright, the authors; Journal compilation copyright, J Clin Med Res and Elmer Press Inc

Journal website https://www.jocmr.org

Short Communication

Volume 15, Number 8-9, September 2023, pages 423-429

Computer-Aided Pulmonary Fibrosis Detection Leveraging an Advanced Artificial Intelligence Triage and Notification Software

Figure 1. ScreenDx-LungFibrosis™ analysis algorithm. (a) ScreenDx-LungFibrosis™ supervised training: The analysis algorithm receives volumetric CT input normalized to standardized pixel values (1) from thousands of labeled cases (i.e., positive or negative for PF). It is then trained (2) using a Convolutional Neural Network (CNN) that incorporates thousands of features via hidden layers (3) into the network. Algorithm output (4) is compared to the “truth”, established by the labeled cases, and feeds back into the model to optimize layer weights. (b) ScreenDx-LungFibrosis™ evaluation of new cases. The analysis algorithm receives volumetric CT input normalized to standardized pixel values (1) from new cases, which proceed through pre-trained hidden layers (2) with fixed weights per layer (3) to create a binary output (4) of 0 (no signs of PF) or 1 (suspicious for PF). PF: pulmonary fibrosis; CT: computed tomography.

**Table 1.** Baseline Cohort Characteristics
Characteristic	PF (n = 692)	No PF (n = 2,326)	Total (n = 3,018)	P value
^aRace and ethnicity data were available in 1,662 (38.5%) of patients. ^bBMI data were available in 52% of patients. ^cOther ILDs include bronchiolitis, chronic hypersensitivity pneumonitis, cryptogenic organizing pneumonia, connective tissue disease-associated ILD, desquamative interstitial pneumonia, eosinophilia granulomatosis with polyangiitis, nonspecific interstitial pneumonia, sarcoidosis, pneumoconiosis, and vasculitis. Note: patients could have multiple primary diagnoses, so column total will not equal 100%. BMI: body mass index; PF: pulmonary fibrosis; IPF: idiopathic pulmonary fibrosis; ILD: interstitial lung disease; CT: computed tomography; ICD: International Classification of Diseases; COVID-19: coronavirus disease 2019; IQR: interquartile range.
Age (years), median (IQR)	69.9 (63.0 - 74.0)	63.2 (58.3 - 71.0)	65.0 (58 - 72)	< 0.001
Age (years), n (%)
≤ 40	1 (0.1)	106 (4.6)	107 (3.5)	-
41 - 50	16 (2.3)	80 (3.4)	96 (3.2)	0.02
51-60	86 (12.4)	591 (25.4)	677 (22.4)	0.69
61-70	254 (36.7)	647 (27.8)	901 (29.8)	< 0.001
> 70	326 (47.1)	453 (19.5)	779 (25.7)	0.06
Male, n (%)	519 (75.0)	1,678 (55.5)	1,678 (55.5)	0.14
Race^a
White	117 (92.1)	1,294 (85.2)	1,411 (85.8)	0.03
Black	15 (3.9)	147 (9.7)	152 (9.2)	0.44
Asian	2 (1.6)	46 (3.0)	48 (2.9)	0.35
Multi	1 (0.8)	22 (1.5)	23 (1.4)	0.54
Hawaiian/Islander	0 (0.0)	8 (0.5)	8 (0.5)	0.41
American Indian	2 (1.6)	1 (< 0.1)	3 (0.2)	< 0.001
Ethnicity^a
Hispanic	14 (10.4)	59 (3.9)	73 (4.4)	< 0.001
Tobacco use, n (%)	469 (67.7)	1,396 (60.0)	1,865 (61.8)	< 0.001
BMI, median (IQR)^b	28.0 (25.4 - 31.4)	27.3 (24.2 - 30.7)	27.3 (24.3 - 30.8)	0.01
Primary diagnoses, n (%)
PF	692 (100.0)	-	692 (22.9)
Normal scan	-	1,072 (35.5)	1,072 (35.5)
Cancer	-	429 (14.2)	429 (14.2)
COVID-19	-	371 (12.3)	371 (12.3)
Emphysema	-	204 (6.8)	204 (6.8)
ILD	692 (100.0)	63 (2.7)	755 (25.1)
IPF	562 (81.2)	0 (0.0)	562 (74.4)
Other ILDs^c	130 (18.8)	63 (100.0)	193 (25.6)
Other	-	93 (4.0)	93 (3.1)
Pneumonia	-	52 (1.7)	52 (1.7)
Granulomatous disease	-	42 (1.4)	42 (1.4)
Method of diagnosis, n (%)
Multi-disciplinary discussion (MDD)	437 (63.2)	533 (22.9)	970 (32.1)
Site-reported diagnosis (clinical or ICD codes)	255 (36.8)	1,793 (77.1)	2,048 (67.9)
CT slice thickness (mm), average (max)	1.6 (5.0)	2.4 (5.0)	2.2 (5.0)	< 0.001

Table 1. Baseline Cohort Characteristics

Characteristic

PF (n = 692)

No PF (n = 2,326)

Total (n = 3,018)

P value

^aRace and ethnicity data were available in 1,662 (38.5%) of patients. ^bBMI data were available in 52% of patients. ^cOther ILDs include bronchiolitis, chronic hypersensitivity pneumonitis, cryptogenic organizing pneumonia, connective tissue disease-associated ILD, desquamative interstitial pneumonia, eosinophilia granulomatosis with polyangiitis, nonspecific interstitial pneumonia, sarcoidosis, pneumoconiosis, and vasculitis. Note: patients could have multiple primary diagnoses, so column total will not equal 100%. BMI: body mass index; PF: pulmonary fibrosis; IPF: idiopathic pulmonary fibrosis; ILD: interstitial lung disease; CT: computed tomography; ICD: International Classification of Diseases; COVID-19: coronavirus disease 2019; IQR: interquartile range.

Age (years), median (IQR)

69.9 (63.0 - 74.0)

63.2 (58.3 - 71.0)

65.0 (58 - 72)

< 0.001

Age (years), n (%)

≤ 40

1 (0.1)

106 (4.6)

107 (3.5)

41 - 50

16 (2.3)

80 (3.4)

96 (3.2)

0.02

51-60

86 (12.4)

591 (25.4)

677 (22.4)

0.69

61-70

254 (36.7)

647 (27.8)

901 (29.8)

< 0.001

> 70

326 (47.1)

453 (19.5)

779 (25.7)

0.06

Male, n (%)

519 (75.0)

1,678 (55.5)

0.14

Race^a

White

117 (92.1)

1,294 (85.2)

1,411 (85.8)

0.03

Black

15 (3.9)

147 (9.7)

152 (9.2)

0.44

Asian

2 (1.6)

46 (3.0)

48 (2.9)

0.35

Multi

1 (0.8)

22 (1.5)

23 (1.4)

0.54

Hawaiian/Islander

0 (0.0)

8 (0.5)

0.41

American Indian

2 (1.6)

1 (< 0.1)

3 (0.2)

< 0.001

Ethnicity^a

Hispanic

14 (10.4)

59 (3.9)

73 (4.4)

< 0.001

Tobacco use, n (%)

469 (67.7)

1,396 (60.0)

1,865 (61.8)

< 0.001

BMI, median (IQR)^b

28.0 (25.4 - 31.4)

27.3 (24.2 - 30.7)

27.3 (24.3 - 30.8)

0.01

Primary diagnoses, n (%)

692 (100.0)

692 (22.9)

Normal scan

1,072 (35.5)

Cancer

429 (14.2)

COVID-19

371 (12.3)

Emphysema

204 (6.8)

ILD

692 (100.0)

63 (2.7)

755 (25.1)

IPF

562 (81.2)

0 (0.0)

562 (74.4)

Other ILDs^c

130 (18.8)

63 (100.0)

193 (25.6)

Other

93 (4.0)

93 (3.1)

Pneumonia

52 (1.7)

Granulomatous disease

42 (1.4)

Method of diagnosis, n (%)

Multi-disciplinary discussion (MDD)

437 (63.2)

533 (22.9)

970 (32.1)

Site-reported diagnosis (clinical or ICD codes)

255 (36.8)

1,793 (77.1)

2,048 (67.9)

CT slice thickness (mm), average (max)

1.6 (5.0)

2.4 (5.0)

2.2 (5.0)

< 0.001

**Table 2.** Diagnostic Parameters of ScreenDx-LungFibrosis™
Test parameter	Value	95% confidence interval
LR: likelihood ratio; OR: odds ratio; PPV: positive predictive value; NPV: negative predictive value.
Sensitivity (%)	91.3	89.0 - 93.3
Specificity (%)	95.1	94.2 - 96.0
LR+	18.8	15.8 - 22.9
LR-	0.09	0.07 - 0.11
OR	206.3	149.0 - 285.6
PPV (%)	84.8	82.1 - 87.3
NPV (%)	97.4	96.6 - 98.0
PPV at low (10%) prevalence of PF (%)	68.0	61.0 - 74.0
NPV at low (10%) prevalence of PF (%)	99.0	98.0 - 100.0
PPV at high (50%) prevalence of PF (%)	95.0	93.0 - 96.0
NPV at high (50%) prevalence of PF (%)	92.0	89.0 - 93.0
Processing time (s), mean	27.6	26.0 - 29.1

Table 2. Diagnostic Parameters of ScreenDx-LungFibrosis™

Test parameter

Value

95% confidence interval

LR: likelihood ratio; OR: odds ratio; PPV: positive predictive value; NPV: negative predictive value.

Sensitivity (%)

91.3

89.0 - 93.3

Specificity (%)

95.1

94.2 - 96.0

LR+

18.8

15.8 - 22.9

LR-

0.09

0.07 - 0.11

206.3

149.0 - 285.6

PPV (%)

84.8

82.1 - 87.3

NPV (%)

97.4

96.6 - 98.0

PPV at low (10%) prevalence of PF (%)

68.0

61.0 - 74.0

NPV at low (10%) prevalence of PF (%)

99.0

98.0 - 100.0

PPV at high (50%) prevalence of PF (%)

95.0

93.0 - 96.0

NPV at high (50%) prevalence of PF (%)

92.0

89.0 - 93.0

Processing time (s), mean

27.6

26.0 - 29.1