Background
Methods
Setting and participants
Lexicon development
Terms and term groups | Uncertainty profiles | ||
---|---|---|---|
Pneumonia-related
|
Non-pneumonia
|
Low uncertainty
|
High uncertainty or Versus
|
Pneumonia | Atelectasis | Probable | Cannot exclude |
Bronchopneumonia | Edema | Consider | Clinical correlation |
Air bronchogram | Congestive heart failure | Concerning for | Could represent |
Consolidation | Heart failure | Consistent with | Possible |
Infiltrate | ARDS | Suspicious | Rule out |
Opacity | Fluid overload | Suspect | Questionable |
Density | Infarct | Suggestive of | Might |
Pneumonitis | Contusion | Likely representing | May |
Pneumonic | Hemorrhage | Compatible with | |
Abcess | Mass | Versus | |
Aspiration | Low lung volume | Plus minus | |
Cavity | Hypoinflation | Or | |
Airspace disease/process | Congestion | And/or | |
Parenchymal process | Malignancy | / | |
Nodule | |||
Neoplasm | |||
Collapse | |||
Effusion | |||
Scar | |||
Fluid |
Natural language processing queries
Physician interpretation
Electronic interpretation
Group (Step) | Determination | Rules | Predicted probability |
---|---|---|---|
Group 1 (Step 1) | Negative | ‘Blanket Negative’ statement without any pneumonia-related terms | |
Group 1 (Step 2) | Negative | No pneumonia-related terms | |
Group 1 (Step 3) | Possible | High uncertainty pneumonia-related terms, no ‘blanket negative’ statement | |
Group 1 (Step 4) | Positive | Low/No uncertainty pneumonia-equivalent terms, no high uncertainty pneumonia-related terms, no non-pneumonia terms | |
Group 2 (Step 5) | Possible | High uncertainty pneumonia-related terms, no low/no uncertainty pneumonia-equivalent terms, no normal statement | |
Group 2 (Step 6) | Possible | Infiltrate + pneumonia-related terms, no low/no uncertainty pneumonia-equivalent terms | |
Group 2 (Step 7) | Possible | Any pneumonia-related versus terms | |
Group 3 (Step 8) | Positive | Low/no uncertainty pneumonia-equivalent terms, no blanket normal statement | |
Group 3 (Step 9) | Possible | Any uncertainty pneumonia-related terms | |
Group 3 (Step 10) | Possible | Any infiltrate + pneumonia-related terms, no non-pneumonia terms | |
Group 4 (Step 11) | Positive | Positive > 70% | |
Group 4 (Step 12) | Negative | Negative >30%, Possible < 30%, Positive < 10% | |
Group 4 (Step 13) | Possible | Possible > 10%, Negative < 10%, Positive < 10% | |
Group 4 (Step 14) | Possible | Possible > 60%, Negative < 40% | |
Group 4 (Step 15) | Positive | Any pneumonia-equivalent term | |
Group 4 (Step 16) | Possible | Possible > 20%, Positive > 10% | |
Group 4 (Step 17) | Possible | Any uncertainty pneumonia-related terms, no low/no uncertainty pneumonia-equivalent terms | Negative < 30% |
Group 4 (Step 18) | Possible | Pneumonia-related terms, no non-pneumonia terms, no blanket normal statement | Negative < 40% |
Group 4 (Step 19) | Negative | Non-pneumonia terms | |
Group 4 (Step 20) | Possible | All remaining reports |
Algorithm performance
Results
Physician interpretation
Positive | |
---|---|
1
|
There is new bilateral lower lobe consolidation with air bronchograms. There is some volume loss. Bibasilar pneumonias.
|
2
|
Again noted is the focal consolidation at the right lung base. It is not significantly changed and most likely represents middle lobe pneumonia. Right middle lobe air space opacity is probably pneumonia and not significantly changed.
|
Possible
| |
3
|
Interval clearing of the diffuse opacities of the lungs with residual opacities, findings suggesting alveolar edema, less likely pneumonia.
|
4
|
Endotracheal tube pulled back. Persistent cardiomegaly with congestive heart failure and bilateral pleural effusions. Bibasilar pneumonia is not excluded.
|
Negative
| |
5
|
Lungs are clear without pulmonary edema, focal consolidation, or pleural effusion. No acute cardiopulmonary disease.
|
6
|
Again seen are diffuse airspace opacities throughout both lungs, improved compared with the most recent prior examination. The pleural effusions appear smaller as well. Persistent pulmonary edema though it appears improved.
|
Clinician interpretation, no. (%) | ||||
---|---|---|---|---|
Sample
|
n
|
Negative
|
Possible
|
Positive
|
Blinded validation |
739
| 424 (57.4) | 262 (35.5) | 53 (7.2) |
Derivation |
950
| 488 (51.4) | 417 (43.9) | 45 (4.7) |
Developmental |
777
| 365 (47.0) | 350 (45.0) | 62 (8.0) |
Overall |
2,466
| 1,277 (51.8) | 1,029 (41.7) | 160 (6.5) |
Lexicon and query development
I2E queries
Electronic algorithm
Test characteristics by interpretation samples (%) | ||||
---|---|---|---|---|
Dataset
|
Sensitivity
|
Specificity
|
PPV
|
NPV
|
Negative-only (versus Positive or Possible)
| ||||
Validation | 92.7 | 91.1 | 93.3 | 90.3 |
Derivation | 93.2 | 96.8 | 96.8 | 93.1 |
Overall | 92.8 | 93.1 | 93.5 | 92.3 |
Positive-only (versus Possible or Negative)
| ||||
Validation | 45.3 | 99.0 | 77.4 | 95.9 |
Derivation | 53.3 | 99.0 | 72.7 | 97.7 |
Overall | 45.0 | 99.0 | 75.8 | 96.3 |
Possible-only (versus Positive or Negative)
| ||||
Validation | 86.6 | 87.4 | 79.1 | 92.3 |
Derivation | 94.2 | 89.9 | 87.9 | 95.2 |
Overall | 89.9 | 87.5 | 83.8 | 92.4 |
Validation set performance
ICU sub-samples
Results by ICU admission diagnosis class | |||
---|---|---|---|
Pneumonia
|
Endocrine/Rheumatologic
| ||
Category | Number (%) | Category | Number (%) |
Incorrect reading
|
7 (4.5)
|
Incorrect reading
| 10 (7.1) |
Normal report | 50 (31.8) | Pneumonia-relevant term | 65 (46.1) |
Heart failure | 34 (21.7) | Atelectasis versus pneumonia-relevant | 40 (28.4) |
Other (e.g., mass, nodules) | 27 (17.2) | Edema versus pneumonia-relevant | 11 (7.8) |
Atelectasis | 16 (6.4) | Other | 8 (5.7) |
Hypoinflation | 10 (4.5) | Pneumonia | 7 (5.0) |
Interstitial markings | 5 (3.2) | ||
Diaphragmatic process | 4 (2.5) | ||
Scar/chronic process | 4 (2.5) |