Multisite implementation of a workflow-integrated machine learning system to optimize COVID-19 hospital admission decisions.

Demand has outstripped healthcare supply during the coronavirus disease 2019 (COVID-19) pandemic. Emergency departments (EDs) are tasked with distinguishing patients who require hospital resources from those who may be safely discharged to the community. The novelty and high variability of COVID-19 have made these determinations challenging. In this study, we developed, implemented and evaluated an electronic health record (EHR) embedded clinical decision support (CDS) system that leverages machine learning (ML) to estimate short-term risk for clinical deterioration in patients with or under investigation for COVID-19. The system translates model-generated risk for critical care needs within 24 h and inpatient care needs within 72 h into rapidly interpretable COVID-19 Deterioration Risk Levels made viewable within ED clinician workflow. ML models were derived in a retrospective cohort of 21,452 ED patients who visited one of five ED study sites and were prospectively validated in 15,670 ED visits that occurred before (n = 4322) or after (n = 11,348) CDS implementation; model performance and numerous patient-oriented outcomes including in-hospital mortality were measured across study periods. Incidence of critical care needs within 24 h and inpatient care needs within 72 h were 10.7% and 22.5%, respectively and were similar across study periods. ML model performance was excellent under all conditions, with AUC ranging from 0.85 to 0.91 for prediction of critical care needs and 0.80-0.90 for inpatient care needs. Total mortality was unchanged across study periods but was reduced among high-risk patients after CDS implementation.

Machine learning and artificial intelligence: applications in healthcare epidemiology.

Artificial intelligence (AI) refers to the performance of tasks by machines ordinarily associated with human intelligence. Machine learning (ML) is a subtype of AI; it refers to the ability of computers to draw conclusions (ie, learn) from data without being directly programmed. ML builds from traditional statistical methods and has drawn significant interest in healthcare epidemiology due to its potential for improving disease prediction and patient care. This review provides an overview of ML in healthcare epidemiology and practical examples of ML tools used to support healthcare decision making at 4 stages of hospital-based care: triage, diagnosis, treatment, and discharge. Examples include model-building efforts to assist emergency department triage, predicting time before septic shock onset, detecting community-acquired pneumonia, and classifying COVID-19 disposition risk level. Increasing availability and quality of electronic health record (EHR) data as well as computing power provides opportunities for ML to increase patient safety, improve the efficiency of clinical management, and reduce healthcare costs.

Abnormal liver chemistries as a predictor of COVID-19 severity and clinical outcomes in hospitalized patients.

BACKGROUND: Abnormal liver chemistries are common findings in patients with Coronavirus Disease 2019 (COVID-19). However, the association of these abnormalities with the severity of COVID-19 and clinical outcomes is poorly understood. AIM: We aimed to assess the prevalence of elevated liver chemistries in hospitalized patients with COVID-19 and compare the serum liver chemistries to predict the severity and in-hospital mortality. METHODS: This retrospective, observational study included 3380 patients with COVID-19 who were hospitalized in the Johns Hopkins Health System (Baltimore, MD, United States). Demographic data, clinical characteristics, laboratory findings, treatment measures, and outcome data were collected. Cox regression modeling was used to explore variables associated with abnormal liver chemistries on admission with disease severity and prognosis. RESULTS: A total of 2698 (70.4%) had abnormal alanine aminotransferase (ALT) at the time of admission. Other more prevalent abnormal liver chemistries were aspartate aminotransferase (AST) (44.4%), alkaline phosphatase (ALP) (16.1%), and total bilirubin (T-Bil) (5.9%). Factors associated with liver injury were older age, Asian ethnicity, other race, being overweight, and obesity. Higher ALT, AST, T-Bil, and ALP levels were more commonly associated with disease severity. Multivariable adjusted Cox regression analysis revealed that abnormal AST and T-Bil were associated with the highest mortality risk than other liver injury indicators during hospitalization. Abnormal AST, T-Bil, and ALP were associated with a need for vasopressor drugs, whereas higher levels of AST, T-Bil, and a decreased albumin levels were associated with mechanical ventilation. CONCLUSION: Abnormal liver chemistries are common at the time of hospital admission in COVID-19 patients and can be closely related to the patient's severity and prognosis. Elevated liver chemistries, specifically ALT, AST, ALP, and T-Bil levels, can be used to stratify risk and predict the need for advanced therapies in these patients.

Liver transplantation in the United States during the COVID-19 pandemic: National and center-level responses.

COVID-19 has profoundly affected the American health care system; its effect on the liver transplant (LT) waitlist based on COVID-19 incidence has not been characterized. Using SRTR data, we compared observed LT waitlist registrations, waitlist mortality, deceased donor LTs (DDLT), and living donor LTs (LDLT) 3/15/2020-8/31/2020 to expected values based on historical trends 1/2016-1/2020, stratified by statewide COVID-19 incidence. Overall, from 3/15 to 4/30, new listings were 11% fewer than expected (IRR = 0.84 0.890.93 ), LDLTs were 49% fewer (IRR = 0.37 0.510.72 ), and DDLTs were 9% fewer (IRR = 0.85 0.910.97 ). In May, new listings were 21% fewer (IRR = 0.74 0.790.84 ), LDLTs were 42% fewer (IRR = 0.39 0.580.85 ) and DDLTs were 13% more (IRR = 1.07 1.151.23 ). Centers in states with the highest incidence 3/15-4/30 had 59% more waitlist deaths (IRR = 1.09 1.592.32 ) and 34% fewer DDLTs (IRR = 0.50 0.660.86 ). By August, waitlist outcomes were occurring at expected rates, except for DDLT (13% more across all incidences). While the early COVID-affected states endured major transplant practice changes, later in the pandemic the newly COVID-affected areas were not impacted to the same extent. These results speak to the adaptability of the transplant community in addressing the pandemic and applying new knowledge to patient care.

Evolving Impact of COVID-19 on Transplant Center Practices and Policies in the United States.

In our first survey of transplant centers in March 2020, >75% of kidney and liver programs were either suspended or operating under restrictions. To safely resume transplantation, we must understand the evolving impact of COVID-19 on transplant recipients and center-level practices. We therefore conducted a six-week follow-up survey May 7-15, 2020, and linked responses to the COVID-19 incidence map, with a response rate of 84%. Suspension of live donor transplantation decreased from 72% in March to 30% in May for kidneys and from 68% to 52% for livers. Restrictions/suspension of deceased donor transplantation decreased from 84% to 58% for kidneys and from 73% to 42% for livers. Resuming transplantation at normal capacity was envisioned by 83% of programs by August 2020. Exclusively using local recovery teams for deceased donor procurement was reported by 28%. Respondents reported caring for a total of 1166 COVID-19-positive transplant recipients; 25% were critically ill. Telemedicine challenges were reported by 81%. There was a lack of consensus regarding management of potential living donors or candidates with SARS-CoV-2. Our findings demonstrate persistent heterogeneity in center-level response to COVID-19 even as transplant activity resumes, making ongoing national data collection and real-time analysis critical to inform best practices.

Impact of the COVID-19 pandemic on commercial airlines in the United States and implications for the kidney transplant community.

Many deceased-donor and living-donor kidney transplants (KTs) rely on commercial airlines for transport. However, the coronavirus-19 pandemic has drastically impacted the commercial airline industry. To understand potential pandemic-related disruptions in the transportation network of kidneys across the United States, we used national flight data to compare scheduled flights during the pandemic vs 1-year earlier, focusing on Organ Procurement Organization (OPO) pairs between which kidneys historically most likely traveled by direct flight (High Volume by direct Air transport OPO Pairs, HVA-OPs). Across the United States, there were 39% fewer flights in April 2020 vs April 2019. Specific to the kidney transportation network, there were 65.1% fewer flights between HVA-OPs, with considerable OPO-level variation (interquartile range [IQR] 54.7%-75.3%; range 0%-100%). This translated to a drop in median number of flights between HVA-OPs from 112 flights/wk in April 2019 to 34 in April 2020 (P < .001), and a rise in wait time between scheduled flights from 1.5 hours in April 2019 (IQR 0.76-3.3) to 4.9 hours in April 2020 (IQR 2.6-11.2; P < .001). Fewer flights and longer wait times can impact logistics as well as cold ischemia time; our findings motivate an exploration of creative approaches to KT transport as the impact of this pandemic on the airline industry evolves.