Study design and ethical considerations
The medical records of previously healthy children admitted for MP infection at our institution between January 2015 and December 2019 were retrospectively reviewed. All methods were carried out in accordance with the format recommended by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. The study protocol was approved by our institutional review board. The review board waived the requirement for informed consent for this study.
Study patients
All patients who had symptoms and signs indicative of pneumonia at admission, including fever (≥38 °C), cough, and abnormal lung auscultation, were included. Empiric antibiotics were initially prescribed for these patients (β-lactam agents and/or macrolides). Only patients initiated on a regimen with macrolides were included. Diagnosis of M. pneumoniae was confirmed by laboratory data and chest radiographs. A baseline blood sample and nasopharyngeal aspirate/swab (NPA) were collected for serological and microbiological testing. M. pneumoniae infection was confirmed using serologic testing and/or polymerase chain reaction (PCR) testing of the NPA. An enzyme immunoassay for IgM antibodies specific to M. pneumoniae (EIA, Bio-Rad PlateliaTM M. pneumoniae IgM, California, USA) was performed with the initial blood samples according to the manufacturer’s protocol.
We excluded patients with underlying diseases, patients who were treated for confirmed or suspected MP infection within the prior four weeks, patients with either positive IgM or PCR for MP but whose symptoms and radiographic findings were incompatible with pneumonia, patients treated with antiviral agents for proven influenza virus with fever onset within 72 hours, patients who received intravenous corticosteroids or changed to alternative antimicrobials (tetracyclines or fluoroquinolones) within 72 hours, and patients who were afebrile after admission.
Definitions of RMPP-3 and RMPP-5
A case with persistent fever for >72 hours without improvement in radiological findings despite appropriate management with macrolides was defined as refractory M. pneumonia pneumonia (RMPP-3). Patients with persistent fever for >120 hours without improvement in radiological findings despite appropriate management were defined as RMPP-5.
Grouping: training and test cohorts for RMPP-3 and RMPP-5
Patients were randomly grouped into the training (n = 501) and test cohorts (n = 215) using simple random sampling without replacement (Figure 1). Each cohort was then categorized into the RMPP-3 group and non-RMPP-3 group based on their duration to defervescence. Defervescence was defined as maintenance of body temperature below 38 °C for at least 24 hours. For the prediction analysis of patients with fever for >120 hours, the group randomization process was implemented again on the same cohort, after which each cohort was categorized into the RMPP-5 and non-RMPP-5 groups.
Predictors: fever profiles
The frequency of fever was defined as the number of peaks on the temperature curve. It was only counted when body temperature was ≥38.0 °C and had increased ≥0.6 °C within 4 hours. If the patient continued to have temperature changes of <0.6 °C but whose body temperature was ≥38.0°C during the 4-hour interval, it was counted as valid (continuous fever pattern).
Predictors: clinical data
Demographic and clinical information were collected in a standardized form by reviewing the electronic medical records. The following information was gathered: duration of fever (before and after hospitalization), total hospital days, and fever profile (highest body temperature, lowest body temperature, frequency of peak fever over 39 °C, frequency of peak fever over 40 °C, and total frequency of peak fever) extracted from 12 sequential fever data within 48 hours. These fever profiles were only included in the analysis for the prediction of prolonged fever over 120 hours (RMPP-5).
Predictors: laboratory data
Tests for complete blood count (CBC), serum aminotransferase, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), lactate dehydrogenase (LDH), procalcitonin, and blood cultures were performed. FilmArray multiplex PCR system (Biomérieux, USA) was used for detecting common respiratory tract virus antigens.
Predictors: radiologic data
Chest radiographs were reviewed independently by two experienced radiologists. They were blinded to the clinical data and original radiographic interpretations. Radiological findings at admission were categorized into four groups: group 1, patients with parahilar peribronchial opacification or diffuse interstitial infiltration; group 2, patients with reticular, nodular, or reticulonodular densities; group 3, patients with segmental or lobar consolidation in a single lobe with or without pleural effusion of 1/4–1/2 in the decubitus position; and group 4, patients with lobar consolidation in 2 or more lobes and/or pleural effusion of more than 1/2 in the decubitus position. The images were interpreted and compared by two radiologists to reach a consensus.
Statistical analyses
Continuous variables were presented as mean ± standard deviation and were compared using an independent t-test. Categorical variables were presented as frequency (%) and were compared using the Pearson chi-squared test or Fisher’s exact test (Table 1 and 2).
Table 1. Initial variables of the whole cohort (fever >72 hours)
|
|
Training cohort (n = 501)
|
Test cohort (n = 215)
|
|
|
Fever ≥72 hours
|
Fever <72 hours
|
p-value
|
Fever ≥72 hours
|
Fever <72 hours
|
p-value
|
Number of patients (n, %)
|
|
163 (32.5)
|
338 (67.5)
|
|
79 (36.7)
|
136 (63.3)
|
|
Age, mean (years)
|
|
5.9 ± 2.8
|
5.5 ± 3.2
|
0.140
|
5.8 ± 2.5
|
5.3 ± 3.1
|
0.273
|
Sex ratio (F:M)
|
|
1.3
|
1.0
|
0.124
|
1.7
|
0.7
|
0.002
|
Duration of fever (days)
|
At admission
|
4.8 ± 1.8
|
5.1 ± 2.2
|
0.057
|
4.9 ± 1.7
|
4.9 ± 2.0
|
0.960
|
|
After admission
|
4.8 ± 3.4
|
1.3 ± 1.0
|
<0.001
|
4.9 ± 3.3
|
1.3 ± 0.9
|
<0.001
|
|
Total
|
9.4 ± 3.3
|
6.5 ± 2.3
|
<0.001
|
9.5 ± 3.0
|
6.3 ± 2.1
|
<0.001
|
Total hospital days (days)
|
|
9.2 ± 3.5
|
5.4 ± 1.6
|
<0.001
|
9.4 ± 3.6
|
5.5 ± 1.8
|
<0.001
|
Initial inflammatory markers
|
WBC × 103/µL
|
7.6 ± 3.5
|
9.6 ± 7.0
|
0.001
|
7.5 ± 3.2
|
9.7 ± 5.5
|
<0.001
|
|
Neutrophils (%)
|
67.0 ± 10.7
|
63.0 ± 14.2
|
0.001
|
68.0 ± 10.8
|
61.9 ± 14.2
|
0.001
|
|
Absolute neutrophil count × 103/µL
|
5.2 ± 2.9
|
6.2 ± 5.3
|
0.003
|
5.1 ± 2.4
|
6.2 ± 4.8
|
0.031
|
|
Lymphocytes (%)
|
23.2 ± 8.4
|
26.6 ± 12.4
|
<0.001
|
22.9 ± 8.4
|
27.7 ± 12.8
|
0.001
|
|
Absolute lymphocyte count × 103/µL
|
1.1 ± 0.5
|
1.4 ± 1.1
|
<0.001
|
1.1 ± 0.6
|
1.4 ± 0.7
|
0.001
|
|
Platelet × 103/µL
|
242.0 ± 75.9
|
304.9 ± 100.4
|
<0.001
|
234.9 ± 71.1
|
308.6 ± 101.1
|
<0.001
|
|
ESR (mm/hr)
|
38.3 ± 18.4
|
35.9 ± 18.5
|
0.178
|
37.4 ± 17.3
|
32.5 ± 17.4
|
<0.001
|
|
CRP (mg/L)
|
49.2 ± 45.6
|
24.7 ± 30.6
|
<0.001
|
57.7 ± 53.9
|
24.1 ± 27.7
|
<0.001
|
|
Procalcitonin (ng/mL)
|
0.6 ± 1.2
|
0.7 ± 2.4
|
0.700
|
0.5 ± 0.7
|
0.3 ± 0.6
|
0.169
|
Other laboratory data
|
LDH (IU/L)
|
359.8 ± 107.4
|
326.2 ± 73.1
|
<0.001
|
372.6 ± 132.8
|
332.7 ± 90.7
|
0.019
|
|
AST (IU/L)
|
40.3 ± 17.9
|
36.2 ± 29.3
|
0.104
|
39.4 ± 14.9
|
41.5 ± 48.5
|
0.701
|
|
ALT (IU/L)
|
19.6 ± 13.3
|
22.6 ± 64.9
|
0.556
|
15.9 ± 7.3
|
29.2 ± 77.9
|
0.051
|
Concurrent respiratory virus (n, %)
|
|
36/140 (25.7)
|
95/285 (33.3)
|
0.110
|
18/64 (28.1)
|
39/117 (33.3)
|
0.471
|
Oxygen requirement (n, %)
|
|
5 (3.1)
|
8 (2.4)
|
0.765
|
3 (3.8)
|
2 (1.5)
|
0.359
|
Radiologic grouping (n, %)
|
Group 1
|
13 (8.0)
|
108 (32.0)
|
<0.001
|
7 (8.9)
|
44 (32.4)
|
<0.001
|
|
Group 2
|
54 (33.1)
|
161 (47.6)
|
22 (27.8)
|
67 (49.3)
|
|
Group 3
|
85 (52.1)
|
68 (20.1)
|
41 (51.9)
|
24 (17.6)
|
|
Group 4
|
11 (6.7)
|
1 (0.3)
|
9 (11.4)
|
1 (0.7)
|
Pleural effusion (n, %)
|
|
34 (20.9)
|
20 (5.9)
|
<0.001
|
18 (22.8)
|
6 (4.4)
|
<0.001
|
Radiologic aggravation on the 3rd or 4th hospital day (n, %)
|
121 (74.2)
|
29 (8.6)
|
<0.001
|
61 (77.2)
|
15 (11.0)
|
<0.001
|
Data are presented as the mean±standard deviation
ESR: Erythrocyte sedimentation rate
CRP: C-reactive protein
LDH: Lactate dehydrogenase
AST: Aspartate aminotransferase
ALT: Alanine aminotransferase
WBC: White blood cell
Table 2. Initial variables of the whole cohort (fever >120 hours)
|
|
Training cohort (n = 501)
|
Test cohort (n = 215)
|
|
|
Fever ≥120 hours
|
Fever <120 hours
|
p-value
|
Fever ≥120 hours
|
Fever <120 hours
|
p-value
|
Number of patients (n, %)
|
|
65 (13.0)
|
436 (87.0)
|
|
31 (14.4)
|
184 (85.6)
|
|
Age, mean (years)
|
|
6.1 ± 2.5
|
5.7 ± 3.1
|
0.217
|
5.7 ± 3.0
|
5.3 ± 3.1
|
0.496
|
Sex ratio (F:M)
|
|
1.8
|
0.9
|
0.011
|
2.4
|
1.0
|
0.030
|
Duration of fever (days)
|
At admission
|
4.2 ± 1.4
|
5.1 ± 2.1
|
<0.001
|
4.4 ± 1.5
|
4.9 ± 2.0
|
0.130
|
|
After admission
|
7.6 ± 3.6
|
1.7 ± 1.2
|
<0.001
|
7.8 ± 4.1
|
1.6 ± 1.2
|
<0.001
|
|
Total
|
11.0 ± 3.8
|
7.0 ± 2.3
|
<0.001
|
11.3 ± 4.6
|
6.6 ± 2.3
|
<0.001
|
Total hospital days (days)
|
|
11.8 ± 3.7
|
6.0 ± 2.0
|
<0.001
|
12.0 ± 4.2
|
5.7 ± 1.7
|
<0.001
|
Initial inflammatory markers
|
WBC × 103/µL
|
7.5 ± 3.4
|
9.4 ± 6.8
|
0.030
|
7.0 ± 3.0
|
8.6 ± 3.7
|
0.031
|
|
Neutrophils (%)
|
71.0 ± 9.6
|
64.0 ± 13.7
|
<0.001
|
66.8 ± 12.1
|
62.0 ± 13.0
|
0.058
|
|
Absolute neutrophil count × 103/µL
|
5.5 ± 3.0
|
6.2 ± 5.2
|
0.290
|
4.5 ± 1.5
|
5.5 ± 3.1
|
0.011
|
|
Lymphocytes (%)
|
20.4 ± 7.2
|
25.8 ± 11.8
|
<0.001
|
23.9 ± 10.0
|
27.5 ± 11.4
|
0.106
|
|
Absolute lymphocyte count × 103/µL
|
1.0 ± 0.4
|
1.3 ± 1.0
|
<0.001
|
1.0 ± 0.5
|
1.3 ± 0.6
|
0.020
|
|
Platelet × 103/µL
|
208.2 ± 50.0
|
298.2 ± 100.9
|
<0.001
|
219.3 ± 72.9
|
286.3 ±89.9
|
<0.001
|
|
ESR (mm/hr)
|
34.2 ± 15.8
|
36.2 ± 18.5
|
0.392
|
32.8 ± 15.6
|
36.5 ± 18.9
|
0.303
|
|
CRP (mg/L)
|
68.4 ± 67.3
|
29.6 ± 33.9
|
<0.001
|
57.9 ± 53.1
|
27.7 ± 25.9
|
0.004
|
|
Procalcitonin (ng/mL)
|
0.7 ± 1.4
|
0.5 ± 1.8
|
0.512
|
1.3 ± 2.3
|
0.5 ± 1.8
|
0.261
|
Other laboratory data
|
LDH (IU/L)
|
379.0 ± 147.6
|
334.4 ± 85.5
|
0.020
|
407.3 ± 139.9
|
328.9 ± 71.5
|
0.005
|
|
AST (IU/L)
|
43.1 ± 22.2
|
38.7 ± 37.3
|
0.346
|
46.2 ± 23.1
|
35.1 ± 11.2
|
0.013
|
|
ALT (IU/L)
|
17.4 ± 11.6
|
25.8 ± 71.6
|
0.351
|
23.2 ± 16.6
|
16.2 ± 10.9
|
0.030
|
Concurrent respiratory virus (n, %)
|
|
13 (24.1)
|
125 (33.7)
|
0.158
|
7 (39.2)
|
43 (27.4)
|
0.856
|
Oxygen requirement (n, %)
|
|
4 (6.2)
|
9 (2.1)
|
0.075
|
2 (6.5)
|
3 (1.6)
|
0.152
|
Radiologic grouping (n, %)
|
Group 1
|
3 (4.6)
|
112 (25.7)
|
<0.001
|
3 (9.7)
|
54 (29.3)
|
<0.001
|
|
Group 2
|
18 (27.7)
|
194 (44.5)
|
6 (19.4)
|
86 (46.7)
|
|
Group 3
|
33 (50.8)
|
126 (28.9)
|
17 (54.8)
|
42 (22.8)
|
|
Group 4
|
11 (16.9)
|
4 (0.9)
|
5 (16.1)
|
2 (1.1)
|
Pleural effusion (n, %)
|
|
19 (29.2)
|
34 (7.8)
|
<0.001
|
9 (29.0)
|
16 (8.7)
|
0.003
|
Radiologic aggravation on the 3rd or 4th hospital day (n, %)
|
|
59 (90.8)
|
108 (24.8)
|
<0.001
|
25 (80.6)
|
34 (18.5)
|
<0.001
|
Fever profiles
|
Highest temperature (˚C)
|
39.6 ± 0.6
|
38.7 ± 0.7
|
<0.001
|
39.4 ± 0.6
|
38.7 ± 0.7
|
<0.001
|
|
Lowest temperature (˚C)
|
37.0 ± 0.3
|
36.6 ± 0.3
|
<0.001
|
36.9 ± 0.3
|
36.6 ± 0.3
|
<0.001
|
|
Frequency of fever >39˚C within 48 hours (n)
|
3.3 ± 2.8
|
0.7 ± 1.4
|
<0.001
|
2.9 ± 1.7
|
0.6 ± 1.2
|
<0.001
|
|
Frequency of fever >40˚C within 48 hours (n)
|
0.5 ± 1.0
|
0.1 ± 0.3
|
<0.001
|
0.3 ± 0.6
|
0.0 ± 0.2
|
0.070
|
|
Frequency of peak fever within 24 hours (n)
|
4.7 ± 1.2
|
2.9 ± 1.4
|
<0.001
|
4.6 ± 1.0
|
2.8 ± 1.3
|
<0.001
|
|
Frequency of peak fever within 48 hours (n)
|
8.7 ± 1.7
|
4.6 ± 2.7
|
<0.001
|
8.6 ± 1.7
|
4.5 ± 2.6
|
<0.001
|
12 sequential body temperatures
|
Initial (˚C)
|
38.6 ± 0.9
|
37.9 ± 0.9
|
<0.001
|
38.2 ± 0.9
|
37.9 ± 0.9
|
0.100
|
|
after 4 hours (˚C)
|
38.1 ± 0.9
|
37.6 ± 0.8
|
<0.001
|
38.0 ± 0.9
|
37.6 ± 0.7
|
0.008
|
|
after 8 hours (˚C)
|
37.9 ± 0.8
|
37.4 ± 0.9
|
<0.001
|
38.0 ± 1.1
|
37.5 ± 0.8
|
0.016
|
|
after 12 hours (˚C)
|
37.9 ± 0.9
|
37.4 ± 0.9
|
<0.001
|
38.0 ± 1.0
|
37.4 ± 0.8
|
<0.001
|
|
after 16 hours (˚C)
|
38.0 ± 1.0
|
37.4 ± 0.8
|
<0.001
|
38.0 ± 0.9
|
37.3 ± 0.8
|
<0.001
|
|
after 20 hours (˚C)
|
38.2 ± 0.9
|
37.4 ± 0.7
|
<0.001
|
38.2 ± 0.6
|
37.4 ± 0.8
|
<0.001
|
|
after 24 hours (˚C)
|
38.0 ± 0.6
|
37.4 ± 0.7
|
<0.001
|
38.2 ± 0.6
|
37.5 ± 0.7
|
<0.001
|
|
after 28 hours (˚C)
|
38.3 ± 1.0
|
37.5 ± 0.8
|
<0.001
|
37.9 ± 0.7
|
37.4 ± 0.7
|
<0.001
|
|
after 32 hours (˚C)
|
37.7 ± 1.0
|
37.0 ± 2.4
|
0.040
|
37.9 ± 1.0
|
37.2 ± 0.7
|
0.001
|
|
after 36 hours (˚C)
|
37.9 ± 0.9
|
37.1 ± 0.7
|
<0.001
|
37.6 ± 0.7
|
37.1 ± 0.8
|
0.001
|
|
after 40 hours (˚C)
|
38.1 ± 0.9
|
37.1 ± 0.7
|
<0.001
|
37.9 ± 0.9
|
37.0 ± 0.7
|
<0.001
|
|
after 44 hours (˚C)
|
38.1 ± 0.8
|
37.1 ± 0.7
|
<0.001
|
37.8 ± 0.8
|
37.1 ± 0.7
|
<0.001
|
Data are presented as the mean±standard deviation
ESR: Erythrocyte sedimentation rate
CRP: C-reactive protein
LDH: Lactate dehydrogenase
AST: Aspartate aminotransferase
ALT: Alanine aminotransferase
WBC: White blood cell
Based on the data from the training cohorts, the univariate logistic regression analysis was performed for identifying significant independent predictors for RMPP-3 or RMPP-5 (Tables 3 and 4). With the significant predictors, stepwise multivariate logistic regression analysis was performed for creating conventional prediction models. To reflect the 12 sequential fever data on the prediction models effectively, a deep neural network (DNN) model was additionally created. DNN included two hidden layers. A dropout layer was used after the first hidden layer to prevent overfitting. For hyperparameter optimization, 20% of the training cohort patients were assigned to the validation cohort. Optimization was performed using the Adam method, and model loss was calculated through binary cross-entropy. Calculations to determine the optimal number of layers and neurons for all DNNs were performed. For each combination of layers and hidden units, hyperparameters for obtaining the best performance for the combination were optimized.
Table 3. Conventional logistic model using data of the training cohort: RMPP-3
|
|
Univariate
|
Multivariate
|
|
|
Odds ratio
|
95% confidence interval
|
p-value
|
Odds ratio
|
95% confidence interval
|
p-value
|
Age (years)
|
|
1.046
|
(0.985, 1.110)
|
0.140
|
|
|
|
Sex (F:M)
|
|
1.345
|
(0.922, 1.955)
|
0.125
|
|
|
|
WBC × 103/µL
|
|
1.000
|
(1.000, 1.000)
|
<0.001
|
|
|
|
Neutrophils (%)
|
|
1.024
|
(1.009, 1.039)
|
0.002
|
|
|
|
Absolute neutrophil count × 103/µL
|
|
1.000
|
(1.000, 1.000)
|
0.018
|
|
|
|
Lymphocytes (%)
|
|
0.972
|
(0.955, 0.990)
|
0.002
|
|
|
|
Absolute lymphocyte count × 103/µL
|
|
0.999
|
(0.999, 1.000)
|
<0.001
|
|
|
|
Hemoglobin (g/dL)
|
|
0.946
|
(0.772, 1.159)
|
0.590
|
|
|
|
Platelets × 103/µL
|
|
0.991
|
(0.989, 0.994)
|
<0.001
|
0.991
|
(0.988, 0.994)
|
<0.001
|
ESR (mm/hr)
|
|
1.007
|
(0.997, 1.017)
|
0.179
|
|
|
|
CRP (mg/L)
|
|
1.019
|
(1.013, 1.026)
|
<0.001
|
1.014
|
(1.008, 1.021)
|
<0.001
|
Procalcitonin (ng/mL)
|
|
0.973
|
(0.845, 1.120)
|
0.700
|
|
|
|
LDH (IU/L)
|
|
1.004
|
(1.002, 1.007)
|
<0.001
|
1.006
|
(1.003, 1.009)
|
<0.001
|
AST (IU/L)
|
|
1.006
|
(0.998, 1.013)
|
0.126
|
|
|
|
ALT (IU/L)
|
|
0.999
|
(0.995, 1.003)
|
0.563
|
|
|
|
Concurrent respiratory virus
|
|
1.444
|
(0.919, 2.270)
|
0.111
|
|
|
|
Oxygen requirement
|
|
0.766
|
(0.247, 2.379)
|
0.645
|
|
|
|
Radiologic grouping
|
|
|
|
<0.001
|
|
|
<0.001
|
|
Group 1
|
-
|
-
|
-
|
|
|
|
|
Group 2
|
2.786
|
(1.451, 5.352)
|
0.002
|
2.408
|
(1.209, 4.796)
|
0.012
|
|
Group 3
|
10.385
|
(5.379, 20.049)
|
<0.001
|
8.080
|
(4.005, 16.302)
|
<0.001
|
|
Group 4
|
91.385
|
(10.899, 766.257)
|
<0.001
|
11.827
|
(1.181, 118.493)
|
0.036
|
Pleural effusion
|
|
4.191
|
(2.325, 7.552)
|
<0.001
|
|
|
|
RMPP-3 : Refractory Mycoplasma pneumoniae pneumonia with fever for ≥ 72 hours
ESR: Erythrocyte sedimentation rate
CRP: C-reactive protein
LDH: Lactate dehydrogenase
AST: Aspartate aminotransferase
ALT: Alanine aminotransferase
Table 4. Conventional logistic model using data of the training cohort: RMPP-5
|
|
Univariate
|
Multivariate
|
|
|
Odds ratio
|
95% confidence interval
|
p-value
|
Odds ratio
|
95% confidence interval
|
p-value
|
Age (years)
|
|
1.046
|
(0.963, 1.138)
|
0.287
|
|
|
|
Sex (F:M)
|
|
0.500
|
(0.292, 0.857)
|
0.012
|
|
|
|
WBC × 103/µL
|
|
1.000
|
(1.000, 1.000)
|
0.017
|
|
|
|
Neutrophils (%)
|
|
1.046
|
(1.022, 1.071)
|
<0.001
|
|
|
|
Absolute neutrophil count × 103/µL
|
|
1.000
|
(1.000, 1.000)
|
0.292
|
|
|
|
Lymphocytes (%)
|
|
0.951
|
(0.924, 0.978)
|
<0.001
|
|
|
|
Absolute lymphocyte count × 103/µL
|
|
0.999
|
(0.998, 0.999)
|
<0.001
|
|
|
|
Hemoglobin (g/dL)
|
|
0.950
|
(0.713, 1.266)
|
0.728
|
|
|
|
Platelet × 103/µL
|
|
0.982
|
(0.977, 0.987)
|
<0.001
|
|
|
|
ESR (mm/hr)
|
|
0.994
|
(0.979, 1.008)
|
0.391
|
|
|
|
CRP (mg/L)
|
|
1.015
|
(1.010, 1.021)
|
<0.001
|
|
|
|
Procalcitonin (ng/mL)
|
|
1.056
|
(0.895, 1.245)
|
0.518
|
|
|
|
LDH (IU/L)
|
|
1.004
|
(1.002, 1.006)
|
0.001
|
|
|
|
AST (IU/L)
|
|
1.003
|
(0.997, 1.009)
|
0.355
|
|
|
|
ALT (IU/L)
|
|
0.996
|
(0.986, 1.006)
|
0.420
|
|
|
|
Concurrent respiratory virus
|
|
0.624
|
(0.323, 1.207)
|
0.161
|
|
|
|
Oxygen requirement
|
|
3.111
|
(0.930, 10.412)
|
0.066
|
|
|
|
Radiologic grouping
|
|
|
|
<0.001
|
|
|
0.007
|
|
Group 1
|
-
|
-
|
-
|
-
|
-
|
-
|
|
Group 2
|
3.464
|
(0.998, 12.020)
|
<0.001
|
1.693
|
(0.447, 6.403)
|
0.438
|
|
Group 3
|
9.778
|
(2.919, 32.757)
|
<0.001
|
3.457
|
(0.942, 12.681)
|
0.061
|
|
Group 4
|
102.667
|
(20.318, 518.780)
|
<0.001
|
13.947
|
(2.260, 86.077)
|
0.005
|
Pleural effusion
|
|
4.884
|
(2.578, 9.252)
|
<0.001
|
|
|
|
Fever profiles
|
Highest temperature (˚C)
|
5.424
|
(3.501, 8.403)
|
<0.001
|
|
|
|
|
Lowest temperature (˚C)
|
54.127
|
(19.031, 153.945)
|
<0.001
|
6.494
|
(2.102, 20.068)
|
0.001
|
|
Frequency of fever of >39˚C within 48 hours (n)
|
1.691
|
(1.494, 1.914)
|
<0.001
|
|
|
|
|
Frequency of fever of >40˚C within 48 hours (n)
|
3.230
|
(2.073, 5.033)
|
<0.001
|
|
|
|
|
Frequency of peak fever within 24 hours (n)
|
2.676
|
(2.097, 3.413)
|
<0.001
|
|
|
|
|
Frequency of peak fever within 48 hours (n)
|
1.872
|
(1.620, 2.162)
|
<0.001
|
1.603
|
(1.361, 1.887)
|
0.001
|
12 sequential body temperatures
|
Initial (˚C)
|
2.017
|
(1.532, 2.654)
|
<0.001
|
|
|
|
|
after 4 hours (˚C)
|
1.988
|
(1.467, 2.695)
|
<0.001
|
|
|
|
|
after 8 hours (˚C)
|
1.852
|
(1.386, 2.474)
|
<0.001
|
|
|
|
|
after 12 hours (˚C)
|
1.960
|
(1.471, 2.610)
|
<0.001
|
|
|
|
|
after 16 hours (˚C)
|
2.134
|
(1.606, 2.834)
|
<0.001
|
|
|
|
|
after 20 hours (˚C)
|
3.740
|
(2.611, 5.358)
|
<0.001
|
|
|
|
|
after 24 hours (˚C)
|
3.138
|
(2.158, 4.562)
|
<0.001
|
|
|
|
|
after 28 hours (˚C)
|
2.926
|
(2.142, 3.996)
|
<0.001
|
|
|
|
|
after 32 hours (˚C)
|
1.930
|
(1.425, 2.615)
|
<0.001
|
|
|
|
|
after 36 hours (˚C)
|
2.929
|
(2.122, 4.043)
|
<0.001
|
|
|
|
|
after 40 hours (˚C)
|
3.446
|
(2.419, 4.908)
|
<0.001
|
|
|
|
|
after 44 hours (˚C)
|
3.873
|
(2.720, 5.515)
|
<0.001
|
|
|
|
RMPP-5 : Refractory Mycoplasma pneumoniae pneumonia with fever for ≥120 hours
ESR: Erythrocyte sedimentation rate
CRP: C-reactive protein
LDH: Lactate dehydrogenase
AST: Aspartate aminotransferase
ALT: Alanine aminotransferase
The prediction power of the conventional logistic prediction model and the DNN model was evaluated in the test cohorts using receiver operating characteristic (ROC) curves (Tables 5 and 6). Logistic regression analysis was performed using SPSS 25 (IBM Corp., Armonk, NY, USA). DNN models were developed using Python 3.7 (open-source projects) with Anaconda 4.7.12, and TensorFlow 2.0.
Table 5. Prediction of RMPP-3 (fever >72 hours) in the test cohort.
Prediction method
|
Used variables
|
p-value
|
AUC area (95% CI)
|
Sensitivity
|
Specificity
|
PPV (Precision)
|
NPV
|
Overall accuracy
|
Youden Index
|
Conventional logistic model
|
CRP, LDH
|
<0.001
|
0.642 (0.562, 0.723)
|
32.9%
|
95.2%
|
81.3%
|
69.4%
|
71.2%
|
0.281
|
Conventional logistic model
|
Radiologic grouping
|
<0.001
|
0.725 (0.651, 0.798)
|
63.3%
|
81.6%
|
66.7%
|
79.3%
|
74.9%
|
0.449
|
Conventional logistic model
|
Platelets, CRP, LDH, radiologic grouping
|
<0.001
|
0.775 (0.704, 0.846)
|
64.6%
|
90.4%
|
79.7%
|
81.5%
|
80.9%
|
0.550
|
Table 6. Prediction of RMPP-5 (fever >120 hours) in the test cohort.
Prediction method
|
Used variables
|
p-value
|
AUC area (95% CI)
|
sensitivity
|
specificity
|
PPV (Precision)
|
NPV
|
Overall accuracy
|
Youden Index
|
Conventional logistic model
|
ALC, CRP, LDH
|
0.251
|
0.565 (0.447, 0.682)
|
12.9%
|
100.0%
|
100.0%
|
87.1%
|
87.4%
|
0.129
|
Conventional logistic model
|
Radiologic grouping
|
0.181
|
0.575 (0.457, 0.693)
|
16.1%
|
98.9%
|
71.4%
|
87.5%
|
87.0%
|
0.150
|
Conventional logistic model
|
ALC, CRP, LDH, Radiologic grouping
|
0.181
|
0.575 (0.457, 0.693)
|
16.1%
|
98.9%
|
71.4%
|
87.5%
|
87.0%
|
0.150
|
Conventional logistic model
|
Fever profiles*
|
0.010
|
0.645 (0.525, 0.765)
|
32.3%
|
96.7%
|
62.5%
|
89.4%
|
87.4%
|
0.290
|
Conventional logistic model
|
Fever profiles*, Radiologic grouping
|
0.019
|
0.632 (0.512, 0.751)
|
29.0%
|
97.3%
|
64.3%
|
89.1%
|
87.4%
|
0.263
|
Deep neural network
|
12 sequential fever data
|
<0.001
|
0.803 (0.699, 0.908)
|
64.5%
|
96.2%
|
74.1%
|
94.1%
|
91.6%
|
0.607
|