Show Your Work: Responsible Model Reporting in Health Care Artificial Intelligence.

Standardized and thorough model reporting is an integral component in the development and deployment of machine learning models in health care. Model reporting includes sharing multiple model performance metrics and incorporating metadata to provide the necessary context for model evaluation. Thorough model reporting addresses common concerns about artificial intelligence in health care including model explainability, transparency, fairness, and generalizability. All stages in the model development lifecycle, from initial design to data capture to model deployment, can be communicated openly to stakeholders with responsible model reporting. Physician involvement throughout these processes can ensure clinical concerns and potential consequences are considered.

Beyond the Spreadsheet: Data Management for Physicians in the Era of Big Data.

Big Data is transforming health care. Characteristics of Big Data require data management strategies to effectively use, analyze, and apply the data. Clinicians are not typically learned in the fundamentals of these strategies which may cause a divide between collected data and data used. This article introduces the fundamentals of Big Data management and encourages clinicians to work with their information technology partners to further understand these processes and to identify opportunities for collaboration.

Ambient air pollution and risk of respiratory infection among adults: evidence from the multiethnic study of atherosclerosis (MESA).

BACKGROUND: Air pollution may affect the risk of respiratory infection, though research has focused on uncommon infections or infections in children. Whether ambient air pollutants increase the risk of common acute respiratory infections among adults is uncertain, yet this may help understand whether pollutants influence spread of pandemic respiratory infections like COVID-19. OBJECTIVE: To estimate the association between ambient air pollutant exposures and respiratory infections in adults. METHODS: During five study examinations over 12 years, 6536 participants in the multiethnic study of atherosclerosis (MESA) reported upper respiratory tract infections, bronchitis, pneumonia or febrile illness in the preceding 2 weeks. Using a validated spatiotemporal model, we estimated residential concentrations of ambient PM2.5, NOx and NO2 for the 2-6 weeks (short-term) and year (long-term) prior to each examination. RESULTS: In this population aged 44-84 years at baseline, 10%-32% of participants reported a recent respiratory infection, depending on month of examination and study region. PM2.5, NOx and NO2 concentrations over the prior 2-6 weeks were associated with increased reporting of recent respiratory infection, with risk ratios (95% CIs) of 1.04 (1.00 to 1.09), 1.15 (1.10 to 1.20) and 1.21 (1.10 to 1.33), respectively, per increase from 25th to 75th percentile in residential pollutant concentration. CONCLUSION: Higher short-term exposure to PM2.5 and traffic-related pollutants are associated with increased risk of symptomatic acute respiratory infections among adults. These findings may provide an insight into the epidemiology of COVID-19.

A Surgeon's Guide to Machine Learning.

Machine learning (ML) represents a collection of advanced data modeling techniques beyond the traditional statistical models and tests with which most clinicians are familiar. While a subset of artificial intelligence, ML is far from the science fiction impression frequently associated with AI. At its most basic, ML is about pattern finding, sometimes with complex algorithms. The advanced mathematical modeling of ML is seeing expanding use throughout healthcare and increasingly in the day-to-day practice of surgeons. As with any new technique or technology, a basic understanding of principles, applications, and limitations are essential for appropriate implementation. This primer is intended to provide the surgical reader an accelerated introduction to applied ML and considerations in potential research applications or the review of publications, including ML techniques.

Big problems in little patients: Nationwide blunt cerebrovascular injury outcomes in the pediatric population.

BACKGROUND: Blunt cerebrovascular injuries (BCVI) are uncommon but potentially devastating. The epidemiology, outcomes, and screening criteria are well described in adults, but data in pediatric patients are extremely limited. The purpose of this study was to characterize pediatric BCVI in a large nationwide sample. We hypothesized that outcomes of BCVI in the pediatric blunt trauma population will vary by age. METHODS: We conducted a retrospective cohort study of the Kids' Inpatient Database for pediatric BCVI from 2000 to 2012. Epidemiology, associated injuries, outcomes (including stroke and mortality), and the utility of standard screening criteria were analyzed. RESULTS: There were 1,182 cases of BCVI identified, yielding an incidence of 0.21%. Patients were predominately male (69%; mean age, 15 ± 5 years). Injuries were 59% carotid, 13% vertebral, and 28% unspecified, with 15% having bilateral or multivessel BCVI. Although younger patients (<11 years) had significantly lower ISS and decreased severe associated injuries (all p < 0.01), they had a similar mortality rate (10%) versus the older cohort. Additionally, the stroke rate was significantly higher among the younger patients versus their older peers (29% mortality for <11 years vs. 15% for ≥11 years, p < 0.01). Only four of seven commonly utilized risk factors were associated with BCVI overall, but none were significantly associated with BCVI in younger children (<11 years). CONCLUSION: This represents the first nationwide assessment of BCVI in the pediatric population. Pediatric BCVI carry considerable mortality and stroke risk. Despite being less severely injured, younger children (<11 years) had similar a mortality rate and a significantly higher stroke rate compared with older pediatric patients. Furthermore, commonly utilized adult screening criteria had limited utility in the younger cohorts. These findings suggest pediatric BCVI may require screening and treatment protocols that are significantly different than currently utilized adult-based programs. LEVEL OF EVIDENCE: Prognostic/Epidemiological Study, level III.

Development and Prospective Validation of a Machine Learning-Based Risk of Readmission Model in a Large Military Hospital.

BACKGROUND: Thirty-day hospital readmissions are a quality metric for health care systems. Predictive models aim to identify patients likely to readmit to more effectively target preventive strategies. Many risk of readmission models have been developed on retrospective data, but prospective validation of readmission models is rare. To the best of our knowledge, none of these developed models have been evaluated or prospectively validated in a military hospital. OBJECTIVES: The objectives of this study are to demonstrate the development and prospective validation of machine learning (ML) risk of readmission models to be utilized by clinical staff at a military medical facility and demonstrate the collaboration between the U.S. Department of Defense's integrated health care system and a private company. METHODS: We evaluated multiple ML algorithms to develop a predictive model for 30-day readmissions using data from a retrospective cohort of all-cause inpatient readmissions at Madigan Army Medical Center (MAMC). This predictive model was then validated on prospective MAMC patient data. Precision, recall, accuracy, and the area under the receiver operating characteristic curve (AUC) were used to evaluate model performance. The model was revised, retrained, and rescored on additional retrospective MAMC data after the prospective model's initial performance was evaluated. RESULTS: Within the initial retrospective cohort, which included 32,659 patient encounters, the model achieved an AUC of 0.68. During prospective scoring, 1,574 patients were scored, of whom 152 were readmitted within 30 days of discharge, with an all-cause readmission rate of 9.7%. The AUC of the prospective predictive model was 0.64. The model achieved an AUC of 0.76 after revision and addition of further retrospective data. CONCLUSION: This work reflects significant collaborative efforts required to operationalize ML models in a complex clinical environment such as that seen in an integrated health care system and the importance of prospective model validation.

Chronic Health Risks in Commercial Fishermen: A Cross-Sectional Analysis from a Small Rural Fishing Village in Alaska.

OBJECTIVES: The purpose of this study is to evaluate chronic health risks before and during the fishing season in a sample of commercial fishermen, addressing the NIOSH priority of Total Worker HealthTM. METHODS: Gillnet license holders in Cordova, Alaska (n = 607) were contacted to participate in a preseason survey (March 2015) assessing health behaviors. A mid-season survey (July 2015) was also conducted. Physical exams and additional assessments were performed on a subset of these fishermen. RESULTS: Sixty-six fishermen participated in the preseason survey and 38 participated in the mid-season survey. The study population was overwhelmingly white males with an average age of 49. The average BMI was 27 with 70% of the participants overweight or obese. Nearly 80% of the sample considered their health good or better. Participants reported longer working hours, less sleep, and less aerobic exercise during the fishing season (P < .05). FitBitTM monitoring (n = 8) confirmed less sleep and fewer steps during fishing season. In one exam (n = 20), 80% of participants showed measured hearing loss at 4 kz (conversation range), and 70% had one or more upper extremity disorders, including 40% with rotator cuff tendonitis. CONCLUSIONS: The prevalence of hearing loss, upper extremity disorders, and sleep apnea risk factors were higher than in the general population both before and during the fishing season. Occupational factors including exposure to noise, the upper extremity demands of gillnetting, and long working hours while fishing exacerbate these chronic health conditions. Health promotion programs targeted toward these conditions may present opportunities for improving total worker health.

Occupational Hydrofluoric Acid Injury from Car and Truck Washing--Washington State, 2001-2013.

Exposure to hydrofluoric acid (HF) causes corrosive chemical burns and potentially fatal systemic toxicity. Car and truck wash cleaning products, rust removers, and aluminum brighteners often contain HF because it is efficient in breaking down roadway matter. The death of a truck wash worker from ingestion of an HF-based wash product and 48 occupational HF burn cases associated with car and truck washing in Washington State during 2001-2013 are summarized in this report. Among seven hospitalized workers, two required surgery, and all but one worker returned to the job. Among 48 injured workers, job titles were primarily auto detailer, car wash worker, truck wash worker, and truck driver. Because HF exposure can result in potentially severe health outcomes, efforts to identify less hazardous alternatives to HF-based industrial wash products are warranted.