Racial Bias in Pulse Oximetry Measurement Sjoding, Michael W; Dickson, Robert P; Iwashyna, Theodore J ...
New England journal of medicine/The New England journal of medicine,
12/2020, Letnik:
383, Številka:
25
Journal Article
Existing prediction models for acute respiratory distress syndrome (ARDS) require manual chart abstraction and have only fair performance-limiting their suitability for driving clinical ...interventions. We sought to develop a machine learning approach for the prediction of ARDS that (a) leverages electronic health record (EHR) data, (b) is fully automated, and (c) can be applied at clinically relevant time points throughout a patient's stay.
We trained a risk stratification model for ARDS using a cohort of 1,621 patients with moderate hypoxia from a single center in 2016, of which 51 patients developed ARDS. We tested the model in a temporally distinct cohort of 1,122 patients from 2017, of which 27 patients developed ARDS. Gold standard diagnosis of ARDS was made by intensive care trained physicians during retrospective chart review. We considered both linear and non-linear approaches to learning the model. The best model used L2-logistic regression with 984 features extracted from the EHR. For patients observed in the hospital at least six hours who then developed moderate hypoxia, the model achieved an area under the receiver operating characteristics curve (AUROC) of 0.81 (95% CI: 0.73-0.88). Selecting a threshold based on the 85th percentile of risk, the model had a sensitivity of 56% (95% CI: 35%, 74%), specificity of 86% (95% CI: 85%, 87%) and positive predictive value of 9% (95% CI: 5%, 14%), identifying a population at four times higher risk for ARDS than other patients with moderate hypoxia and 17 times the risk of hospitalized adults.
We developed an ARDS prediction model based on EHR data with good discriminative performance. Our results demonstrate the feasibility of a machine learning approach to risk stratifying patients for ARDS solely from data extracted automatically from the EHR.
Abstract
Objective
In applying machine learning (ML) to electronic health record (EHR) data, many decisions must be made before any ML is applied; such preprocessing requires substantial effort and ...can be labor-intensive. As the role of ML in health care grows, there is an increasing need for systematic and reproducible preprocessing techniques for EHR data. Thus, we developed FIDDLE (Flexible Data-Driven Pipeline), an open-source framework that streamlines the preprocessing of data extracted from the EHR.
Materials and Methods
Largely data-driven, FIDDLE systematically transforms structured EHR data into feature vectors, limiting the number of decisions a user must make while incorporating good practices from the literature. To demonstrate its utility and flexibility, we conducted a proof-of-concept experiment in which we applied FIDDLE to 2 publicly available EHR data sets collected from intensive care units: MIMIC-III and the eICU Collaborative Research Database. We trained different ML models to predict 3 clinically important outcomes: in-hospital mortality, acute respiratory failure, and shock. We evaluated models using the area under the receiver operating characteristics curve (AUROC), and compared it to several baselines.
Results
Across tasks, FIDDLE extracted 2,528 to 7,403 features from MIMIC-III and eICU, respectively. On all tasks, FIDDLE-based models achieved good discriminative performance, with AUROCs of 0.757–0.886, comparable to the performance of MIMIC-Extract, a preprocessing pipeline designed specifically for MIMIC-III. Furthermore, our results showed that FIDDLE is generalizable across different prediction times, ML algorithms, and data sets, while being relatively robust to different settings of user-defined arguments.
Conclusions
FIDDLE, an open-source preprocessing pipeline, facilitates applying ML to structured EHR data. By accelerating and standardizing labor-intensive preprocessing, FIDDLE can help stimulate progress in building clinically useful ML tools for EHR data.
When hospitals are at capacity, accurate deterioration indices could help identify low-risk patients as potential candidates for home care programs and alleviate hospital strain. To date, many ...existing deterioration indices are based entirely on structured data from the electronic health record (EHR) and ignore potentially useful information from other sources.
To improve the accuracy of existing deterioration indices by incorporating unstructured imaging data from chest radiographs.
Machine learning models were trained to predict deterioration of patients hospitalized with acute dyspnea using existing deterioration index scores and chest radiographs. Models were trained on hospitalized patients without coronavirus disease 2019 (COVID-19) and then subsequently tested on patients with COVID-19 between January 2020 and December 2020 at a single tertiary care center who had at least one radiograph taken within 48 hours of hospital admission.
Patient deterioration was defined as the need for invasive or non-invasive mechanical ventilation, heated high flow nasal cannula, IV vasopressor administration or in-hospital mortality at any time following admission. The EPIC deterioration index was augmented with unstructured data from chest radiographs to predict risk of deterioration. We compared discriminative performance of the models with and without incorporating chest radiographs using area under the receiver operating curve (AUROC), focusing on comparing the fraction and total patients identified as low risk at different negative predictive values (NPV).
Data from 6278 hospitalizations were analyzed, including 5562 hospitalizations without COVID-19 (training cohort) and 716 with COVID-19 (216 in validation, 500 in held-out test cohort). At a NPV of 0.95, the best-performing image-augmented deterioration index identified 49 more (9.8%) individuals as low-risk compared to the deterioration index based on clinical data alone in the first 48 hours of admission. At a NPV of 0.9, the EPIC image-augmented deterioration index identified 26 more individuals (5.2%) as low-risk compared to the deterioration index based on clinical data alone in the first 48 hours of admission.
Augmenting existing deterioration indices with chest radiographs results in better identification of low-risk patients. The model augmentation strategy could be used in the future to incorporate other forms of unstructured data into existing disease models.
Recent studies have revealed that, in critically ill patients, lung microbiota are altered and correlate with alveolar inflammation. The clinical significance of altered lung bacteria in critical ...illness is unknown.
To determine if clinical outcomes of critically ill patients are predicted by features of the lung microbiome at the time of admission.
We performed a prospective, observational cohort study in an ICU at a university hospital. Lung microbiota were quantified and characterized using droplet digital PCR and bacterial 16S ribosomal RNA gene quantification and sequencing. Primary predictors were the bacterial burden, community diversity, and community composition of lung microbiota. The primary outcome was ventilator-free days, determined at 28 days after admission.
Lungs of 91 critically ill patients were sampled using miniature BAL within 24 hours of ICU admission. Patients with increased lung bacterial burden had fewer ventilator-free days (hazard ratio, 0.43; 95% confidence interval, 0.21-0.88), which remained significant when the analysis was controlled for pneumonia and severity of illness. The community composition of lung bacteria predicted ventilator-free days (
= 0.003), driven by the presence of gut-associated bacteria (e.g., species of the Lachnospiraceae and Enterobacteriaceae families). Detection of gut-associated bacteria was also associated with the presence of acute respiratory distress syndrome.
Key features of the lung microbiome (bacterial burden and enrichment with gut-associated bacteria) predict outcomes in critically ill patients. The lung microbiome is an understudied source of clinical variation in critical illness and represents a novel therapeutic target for the prevention and treatment of acute respiratory failure.
Failure to reliably diagnose ARDS may be a major driver of negative clinical trials and underrecognition and treatment in clinical practice. We sought to examine the interobserver reliability of the ...Berlin ARDS definition and examine strategies for improving the reliability of ARDS diagnosis.
Two hundred five patients with hypoxic respiratory failure from four ICUs were reviewed independently by three clinicians, who evaluated whether patients had ARDS, the diagnostic confidence of the reviewers, whether patients met individual ARDS criteria, and the time when criteria were met.
Interobserver reliability of an ARDS diagnosis was “moderate” (kappa = 0.50; 95% CI, 0.40-0.59). Sixty-seven percent of diagnostic disagreements between clinicians reviewing the same patient was explained by differences in how chest imaging studies were interpreted, with other ARDS criteria contributing less (identification of ARDS risk factor, 15%; cardiac edema/volume overload exclusion, 7%). Combining the independent reviews of three clinicians can increase reliability to “substantial” (kappa = 0.75; 95% CI, 0.68-0.80). When a clinician diagnosed ARDS with “high confidence,” all other clinicians agreed with the diagnosis in 72% of reviews. There was close agreement between clinicians about the time when a patient met all ARDS criteria if ARDS developed within the first 48 hours of hospitalization (median difference, 5 hours).
The reliability of the Berlin ARDS definition is moderate, driven primarily by differences in chest imaging interpretation. Combining independent reviews by multiple clinicians or improving methods to identify bilateral infiltrates on chest imaging are important strategies for improving the reliability of ARDS diagnosis.
Acute respiratory distress syndrome (ARDS) is an acute inflammatory lung injury characterized by severe hypoxemic respiratory failure, bilateral opacities on chest imaging, and low lung compliance. ...ARDS is a heterogeneous syndrome that is the common end point of a wide variety of predisposing conditions, with complex pathophysiology and underlying mechanisms. Routine management of ARDS is centered on lung-protective ventilation strategies such as low tidal volume ventilation and targeting low airway pressures to avoid exacerbation of lung injury, as well as a conservative fluid management strategy.
Understanding risk factors for short- and long-term COVID-19 outcomes have implications for current guidelines and practice. We study whether early identified risk factors for COVID-19 persist one ...year later and through varying disease progression trajectories.
This was a retrospective study of 6,731 COVID-19 patients presenting to Michigan Medicine between March 10, 2020 and March 10, 2021. We describe disease progression trajectories from diagnosis to potential hospital admission, discharge, readmission, or death. Outcomes pertained to all patients: rate of medical encounters, hospitalization-free survival, and overall survival, and hospitalized patients: discharge versus in-hospital death and readmission. Risk factors included patient age, sex, race, body mass index, and 29 comorbidity conditions.
Younger, non-Black patients utilized healthcare resources at higher rates, while older, male, and Black patients had higher rates of hospitalization and mortality. Diabetes with complications, coagulopathy, fluid and electrolyte disorders, and blood loss anemia were risk factors for these outcomes. Diabetes with complications, coagulopathy, fluid and electrolyte disorders, and blood loss were associated with lower discharge and higher inpatient mortality rates.
This study found differences in healthcare utilization and adverse COVID-19 outcomes, as well as differing risk factors for short- and long-term outcomes throughout disease progression. These findings may inform providers in emergency departments or critical care settings of treatment priorities, empower healthcare stakeholders with effective disease management strategies, and aid health policy makers in optimizing allocations of medical resources.
Acute hypoxemic respiratory failure (AHRF) and acute respiratory distress syndrome (ARDS) are associated with high in-hospital mortality. However, in cohorts of ARDS patients from the 1990s, patients ...more commonly died from sepsis or multi-organ failure rather than refractory hypoxemia. Given increased attention to lung-protective ventilation and sepsis treatment in the past 25 years, we hypothesized that causes of death may be different among contemporary cohorts. These differences may provide clinicians with insight into targets for future therapeutic interventions.
We identified adult patients hospitalized at a single tertiary care center (2016-2017) with AHRF, defined as PaO
/FiO
≤ 300 while receiving invasive mechanical ventilation for > 12 h, who died during hospitalization. ARDS was adjudicated by multiple physicians using the Berlin definition. Separate abstractors blinded to ARDS status collected data on organ dysfunction and withdrawal of life support using a standardized tool. The primary cause of death was defined as the organ system that most directly contributed to death or withdrawal of life support.
We identified 385 decedents with AHRF, of whom 127 (33%) had ARDS. The most common primary causes of death were sepsis (26%), pulmonary dysfunction (22%), and neurologic dysfunction (19%). Multi-organ failure was present in 70% at time of death, most commonly due to sepsis (50% of all patients), and 70% were on significant respiratory support at the time of death. Only 2% of patients had insupportable oxygenation or ventilation. Eighty-five percent died following withdrawal of life support. Patients with ARDS more often had pulmonary dysfunction as the primary cause of death (28% vs 19%; p = 0.04) and were also more likely to die while requiring significant respiratory support (82% vs 64%; p < 0.01).
In this contemporary cohort of patients with AHRF, the most common primary causes of death were sepsis and pulmonary dysfunction, but few patients had insupportable oxygenation or ventilation. The vast majority of deaths occurred after withdrawal of life support. ARDS patients were more likely to have pulmonary dysfunction as the primary cause of death and die while requiring significant respiratory support compared to patients without ARDS.