Diagnostic Uncertainty in Dyspneic Patients with Cancer in the Emergency Department

A novel interpretable deep learning model for diagnosis in emergency department dyspnoea patients based on complete data from an entire health care system

BACKGROUND

Dyspnoea is one of the emergency department's (ED) most common and deadly chief complaints, but frequently misdiagnosed and mistreated. We aimed to design a diagnostic decision support which classifies dyspnoeic ED visits into acute heart failure (AHF), exacerbation of chronic obstructive pulmonary disease (eCOPD), pneumonia and "other diagnoses" by using deep learning and complete, unselected data from an entire regional health care system.

METHODS

In this cross-sectional study, we included all dyspnoeic ED visits of patients ≥ 18 years of age at the two EDs in the region of Halland, Sweden, 07/01/2017-12/31/2019. Data from the complete regional health care system within five years prior to the ED visit were analysed. Gold standard diagnoses were defined as the subsequent in-hospital or ED discharge notes, and a subsample was manually reviewed by emergency medicine experts. A novel deep learning model, the clinical attention-based recurrent encoder network (CareNet), was developed. Cohort performance was compared to a simpler CatBoost model. A list of all variables and their importance for diagnosis was created. For each unique patient visit, the model selected the most important variables, analysed them and presented them to the clinician interpretably by taking event time and clinical context into account. AUROC, sensitivity and specificity were compared.

FINDINGS

The most prevalent diagnoses among the 10,315 dyspnoeic ED visits were AHF (15.5%), eCOPD (14.0%) and pneumonia (13.3%). Median number of unique events, i.e., registered clinical data with time stamps, per ED visit was 1,095 (IQR 459-2,310). CareNet median AUROC was 87.0%, substantially higher than the CatBoost model´s (81.4%). CareNet median sensitivity for AHF, eCOPD, and pneumonia was 74.5%, 92.6%, and 54.1%, respectively, with a specificity set above 75.0, slightly inferior to that of the CatBoost baseline model. The model assembled a list of 1,596 variables by importance for diagnosis, on top were prior diagnoses of heart failure or COPD, daily smoking, atrial fibrillation/flutter, life management difficulties and maternity care. Each patient visit received their own unique attention plot, graphically displaying important clinical events for the diagnosis.

INTERPRETATION

We designed a novel interpretable deep learning model for diagnosis in emergency department dyspnoea patients by analysing unselected data from a complete regional health care system.

Characterization of older adults with cancer seeking acute emergency department care: A prospective observational study

INTRODUCTION

Disparities in care of older adults in cancer treatment trials and emergency department (ED) use exist. This report provides a baseline description of older adults ≥65 years old who present to the ED with active cancer.

MATERIALS AND METHODS

Planned secondary analysis of the Comprehensive Oncologic Emergencies Research Network observational ED cohort study sponsored by the National Cancer Institute. Of 1564 eligible adults with active cancer, 1075 patients were prospectively enrolled, of which 505 were ≥ 65 years old. We recruited this convenience sample from eighteen participating sites across the United States between February 1, 2016 and January 30, 2017.

RESULTS

Compared to cancer patients younger than 65 years of age, older adults were more likely to be transported to the ED by emergency medical services, have a higher Charlson Comorbidity Index score, and be admitted despite no significant difference in acuity as measured by the Emergency Severity Index. Despite the higher admission rate, no significant difference was noted in hospitalization length of stay, 30-day mortality, ED revisit or hospital admission within 30 days after the index visit. Three of the top five ED diagnoses for older adults were symptom-related (fever of other and unknown origin, abdominal and pelvic pain, and pain in throat and chest). Despite this, older adults were less likely to report symptoms and less likely to receive symptomatic treatment for pain and nausea than the younger comparison group. Both younger and older adults reported a higher symptom burden on the patient reported Condensed Memorial Symptom Assessment Scale than to ED providers. When treating suspected infection, no differences were noted in regard to administration of antibiotics in the ED, admissions, or length of stay ≤2 days for those receiving ED antibiotics.

DISCUSSION

We identified several differences between older (≥65 years old) and younger adults with active cancer seeking emergency care. Older adults frequently presented for symptom-related diagnoses but received fewer symptomatic interventions in the ED suggesting that important opportunities to improve the care of older adults with cancer in the ED exist.

Risk factors and predictors of immune-related adverse events: implications for patients with non-small cell lung cancer

INTRODUCTION

Immune checkpoint inhibitors (ICI) are now utilized as a standard of care treatment for multiple cancers, including in both the metastatic setting as well as in earlier stages of disease. The identification of unique immune-related adverse events (irAE) that occur during ICI treatment has led to intense research to identify potential risk factors and biomarkers that may assist in clinical decision making. Although initial studies in ICI were primarily in advanced stage disease, the use of ICI in earlier stages of disease as adjuvant therapies requires a better understanding of patient risk stratification to mitigate or prevent serious irAE.

AREAS COVERED

In this review, we set out to describe the current state of research regarding potential risk factors for irAE in patients with non-small cell lung cancer, as well as explore the barriers to understanding irAE. We review data from irAE that occur in large phase 3 trials and prospective studies focusing on irAE, as well as the many retrospective studies that currently form the bulk of our understanding of irAE.