Open-source machine learning pipeline automatically flags instances of acute respiratory distress syndrome from electronic health records.

Physicians could greatly benefit from automated diagnosis and prognosis tools to help address information overload and decision fatigue. Intensive care physicians stand to benefit greatly from such tools as they are at particularly high risk for those factors. Acute Respiratory Distress Syndrome (ARDS) is a life-threatening condition affecting >10% of critical care patients and has a mortality rate over 40%. However, recognition rates for ARDS have been shown to be low (30-70%) in clinical settings. In this work, we present a reproducible computational pipeline that automatically adjudicates ARDS on retrospective datasets of mechanically ventilated adult patients. This pipeline automates the steps outlined by the Berlin Definition through implementation of natural language processing tools and classification algorithms. We train an XGBoost model on chest imaging reports to detect bilateral infiltrates, and another on a subset of attending physician notes labeled for the most common ARDS risk factor in our data. Both models achieve high performance-a minimum area under the receiver operating characteristic curve (AUROC) of 0.86 for adjudicating chest imaging reports in out-of-bag test sets, and an out-of-bag AUROC of 0.85 for detecting a diagnosis of pneumonia. We validate the entire pipeline on a cohort of MIMIC-III encounters and find a sensitivity of 93.5% - an extraordinary improvement over the 22.6% ARDS recognition rate reported for these encounters - along with a specificity of 73.9%. We conclude that our reproducible, automated diagnostic pipeline exhibits promising accuracy, generalizability, and probability calibration, thus providing a valuable resource for physicians aiming to enhance ARDS diagnosis and treatment strategies. We surmise that proper implementation of the pipeline has the potential to aid clinical practice by facilitating the recognition of ARDS cases at scale.

Validation of a tool for estimating clinician recognition of ARDS using data from the international LUNG SAFE study.

Under-recognition of acute respiratory distress syndrome (ARDS) by clinicians is an important barrier to adoption of evidence-based practices such as low tidal volume ventilation. The burden created by the COVID-19 pandemic makes it even more critical to develop scalable data-driven tools to improve ARDS recognition. The objective of this study was to validate a tool for accurately estimating clinician ARDS recognition rates using discrete clinical characteristics easily available in electronic health records. We conducted a secondary analysis of 2,705 ARDS and 1,261 non-ARDS hypoxemic patients in the international LUNG SAFE cohort. The primary outcome was validation of a tool that estimates clinician ARDS recognition rates from health record data. Secondary outcomes included the relative impact of clinical characteristics on tidal volume delivery and clinician documentation of ARDS. In both ARDS and non-ARDS patients, greater height was associated with lower standardized tidal volume (mL/kg PBW) (ARDS: adjusted ß = -4.1, 95% CI -4.5 --3.6; non-ARDS: ß = -7.7, 95% CI -8.8 --6.7, P<0.00009 [where α = 0.01/111 with the Bonferroni correction]). Standardized tidal volume has already been normalized for patient height, and furthermore, height was not associated with clinician documentation of ARDS. Worsening hypoxemia was associated with both increased clinician documentation of ARDS (ß = -0.074, 95% CI -0.093 --0.056, P<0.00009) and lower standardized tidal volume (ß = 1.3, 95% CI 0.94-1.6, P<0.00009) in ARDS patients. Increasing chest imaging opacities, plateau pressure, and clinician documentation of ARDS also were associated with lower tidal volume in ARDS patients. Our EHR-based data-driven approach using height, gender, ARDS documentation, and lowest standardized tidal volume yielded estimates of clinician ARDS recognition rates of 54% for mild, 63% for moderate, and 73% for severe ARDS. Our tool replicated clinician-reported ARDS recognition in the LUNG SAFE study, enabling the identification of ARDS patients at high risk of being unrecognized. Our approach can be generalized to other conditions for which there is a need to increase adoption of evidence-based care.

Chest radiograph at admission predicts early intubation among inpatient COVID-19 patients.

OBJECTIVE: The 2019 Coronavirus (COVID-19) results in a wide range of clinical severity and there remains a need for prognostic tools which identify patients at risk of rapid deterioration and who require critical care. Chest radiography (CXR) is routinely obtained at admission of COVID-19 patients. However, little is known regarding correlates between CXR severity and time to intubation. We hypothesize that the degree of opacification on CXR at time of admission independently predicts need and time to intubation. METHODS: In this retrospective cohort study, we reviewed COVID-19 patients who were admitted to an urban medical center during March 2020 that had a CXR performed on the day of admission. CXRs were divided into 12 lung zones and were assessed by two blinded thoracic radiologists. A COVID-19 opacification rating score (CORS) was generated by assigning one point for each lung zone in which an opacity was observed. Underlying comorbidities were abstracted and assessed for association. RESULTS: One hundred forty patients were included in this study and 47 (34%) patients required intubation during the admission. Patients with CORS ≥ 6 demonstrated significantly higher rates of early intubation within 48 h of admission and during the hospital stay (ORs 24 h, 19.8, p < 0.001; 48 h, 28.1, p < 0.001; intubation during hospital stay, 6.1, p < 0.0001). There was no significant correlation between CORS ≥ 6 and age, sex, BMI, or any underlying cardiac or pulmonary comorbidities. CONCLUSIONS: CORS ≥ 6 at the time of admission predicts need for intubation, with significant increases in intubation at 24 and 48 h, independent of comorbidities. KEY POINTS: • Chest radiography at the time of admission independently predicts time to intubation within 48 h and during the hospital stay in COVID-19 patients. • More opacities on chest radiography are associated with several fold increases in early mechanical ventilation among COVID-19 patients. • Chest radiography is useful in identifying COVID-19 patients whom may rapidly deteriorate and help inform clinical management as well as hospital bed and ventilation allocation.

A quantitative approach for the analysis of clinician recognition of acute respiratory distress syndrome using electronic health record data.

IMPORTANCE: Despite its efficacy, low tidal volume ventilation (LTVV) remains severely underutilized for patients with acute respiratory distress syndrome (ARDS). Physician under-recognition of ARDS is a significant barrier to LTVV use. We propose a computational method that addresses some of the limitations of the current approaches to automated measurement of whether ARDS is recognized by physicians. OBJECTIVE: To quantify patient and physician factors affecting physicians' tidal volume selection and to build a computational model of physician recognition of ARDS that accounts for these factors. DESIGN, SETTING, AND PARTICIPANTS: In this cross-sectional study, electronic health record data were collected for 361 ARDS patients and 388 non-ARDS hypoxemic (control) patients in nine adult intensive care units at four hospitals between June 24 and December 31, 2013. METHODS: Standardized tidal volumes (mL/kg predicted body weight) were chosen as a proxy for physician decision-making behavior. Using data-science approaches, we quantified the effect of eight factors (six severity of illness, two physician behaviors) on selected standardized tidal volumes in ARDS and control patients. Significant factors were incorporated in computational behavioral models of physician recognition of ARDS. RESULTS: Hypoxemia severity and ARDS documentation in physicians' notes were associated with lower standardized tidal volumes in the ARDS cohort. Greater patient height was associated with lower standardized tidal volumes (which is already normalized for height) in both ARDS and control patients. The recognition model yielded a mean (99% confidence interval) physician recognition of ARDS of 22% (9%-42%) for mild, 34% (19%-49%) for moderate, and 67% (41%-100%) for severe ARDS. CONCLUSIONS AND RELEVANCE: In this study, patient characteristics and physician behaviors were demonstrated to be associated with differences in ventilator management in both ARDS and control patients. Our model of physician ARDS recognition measurement accounts for these clinical variables, providing an electronic approach that moves beyond relying on chart documentation or resource intensive approaches.