RESUMEN
Background: Fatal coronary heart disease (FCHD) is often described as sudden cardiac death (affects >4 million people/year), where coronary artery disease is the only identified condition. Electrocardiographic artificial intelligence (ECG-AI) models for FCHD risk prediction using ECG data from wearable devices could enable wider screening/monitoring efforts. Objectives: To develop a single-lead ECG-based deep learning model for FCHD risk prediction and assess concordance between clinical and Apple Watch ECGs. Methods: An FCHD single-lead ("lead I" from 12-lead ECGs) ECG-AI model was developed using 167,662 ECGs (50,132 patients) from the University of Tennessee Health Sciences Center. Eighty percent of the data (5-fold cross-validation) was used for training and 20% as a holdout. Cox proportional hazards (CPH) models incorporating ECG-AI predictions with age, sex, and race were also developed. The models were tested on paired clinical single-lead and Apple Watch ECGs from 243 St. Jude Lifetime Cohort Study participants. The correlation and concordance of the predictions were assessed using Pearson correlation (R), Spearman correlation (ρ), and Cohen's kappa. Results: The ECG-AI and CPH models resulted in AUC = 0.76 and 0.79, respectively, on the 20% holdout and AUC = 0.85 and 0.87 on the Atrium Health Wake Forest Baptist external validation data. There was moderate-strong positive correlation between predictions (R = 0.74, ρ = 0.67, and κ = 0.58) when tested on the 243 paired ECGs. The clinical (lead I) and Apple Watch predictions led to the same low/high-risk FCHD classification for 99% of the participants. CPH prediction correlation resulted in an R = 0.81, ρ = 0.76, and κ = 0.78. Conclusion: Risk of FCHD can be predicted from single-lead ECGs obtained from wearable devices and are statistically concordant with lead I of a 12-lead ECG.
RESUMEN
OBJECTIVE: We sought to examine outcomes of ultrafiltration in real world community-based hospital settings. BACKGROUND: Ultrafiltration (UF) is an accepted therapeutic option for advanced decompensated heart failure (ADHF). the feasibility of UF in a community hospital setting, by general cardiologists in a start-up program had not been objectively evaluated. METHODS: We retrospectively analyzed the first-year cohort of ADHF patients treated with UF from 10/1/2019 to 10/1/2020, which totaled 30 patients, utilizing the CHF Solutions Aquadex FlexFlow™ System with active UF rate titration. RESULTS: Baseline patient characteristics were similar to RCTs: mean age 63, 73 % male; 27 % female; 53 % Caucasian; 47 % African American; 77 % had LVEF ≤ 40. The baseline mean serum creatinine (Cr) was 1.84 ±0.62 mg/dL, mean GFR of 36.95 ±9.60 ml/min. HF re-admission rates were not significantly different than prior studies (17.2 % at 30 d, 23.3 % at 60 d, but in our cohort, per patient HF re-admission rates were reduced significantly by 60 d (0.30 p = 0.017). CONCLUSION: Our analysis showed success with UF in mainstream setting with reproducible results of significant volume loss without adverse renal effect, mitigation of recurrent Hdmissions, and remarkable subjective clinical benefit.
RESUMEN
Introduction: More than 76,000 women die yearly from preeclampsia and hypertensive disorders of pregnancy. Early diagnosis and management of preeclampsia can improve outcomes for both mother and baby. In this study, we developed artificial intelligence models to detect and predict preeclampsia from electrocardiograms (ECGs) in point-of-care settings. Methods: Ten-second 12-lead ECG data was obtained from two large health care settings: University of Tennessee Health Science Center (UTHSC) and Atrium Health Wake Forest Baptist (AHWFB). UTHSC data was split into 80% training and 20% holdout data. The model used a modified ResNet convolutional neural network, taking one-dimensional raw ECG signals comprising 12 channels as an input, to predict risk of preeclampsia. Sub-analyses were performed to assess the predictive accuracy for preeclampsia prediction within 30, 60, or 90 days before diagnosis. Results: The UTHSC cohort included 904 ECGs from 759 females (78.8% African American) with a mean ± sd age of 27.3 ± 5.0 years. The AHWFB cohort included 817 ECGs from 141 females (45.4 African American) with a mean ± sd age of 27.4 ± 5.9 years. The cross-validated ECG-AI model yielded an AUC (95% CI) of 0.85 (0.77-0.93) on UTHSC holdout data, and an AUC (95% CI) of 0.81 (0.77-0.84) on AHWFB data. The sub-analysis of different time windows before preeclampsia prediction resulted in AUCs (95% CI) of 0.92 (0.84-1.00), 0.89 (0.81-0.98) and 0.90 (0.81-0.98) when tested on ECGs 30 days, 60 days and 90 days, respectively, before diagnosis. When assessed on early onset preeclampsia (preeclampsia diagnosed at <34 weeks of pregnancy), the model's AUC (95% CI) was 0.98 (0.89-1.00). Discussion: We conclude that preeclampsia can be identified with high accuracy via application of AI models to ECG data.
RESUMEN
BACKGROUND: This study used electrocardiogram data in conjunction with artificial intelligence methods as a noninvasive tool for detecting peripartum cardiomyopathy. OBJECTIVE: This study aimed to assess the efficacy of an artificial intelligence-based heart failure detection model for peripartum cardiomyopathy detection. STUDY DESIGN: We first built a deep-learning model for heart failure detection using retrospective data at the University of Tennessee Health Science Center. Cases were adult and nonpregnant female patients with a heart failure diagnosis; controls were adult nonpregnant female patients without heart failure. The model was then tested on an independent cohort of pregnant women at the University of Tennessee Health Science Center with or without peripartum cardiomyopathy. We also tested the model in an external cohort of pregnant women at Atrium Health Wake Forest Baptist. Key outcomes were assessed using the area under the receiver operating characteristic curve. We also repeated our analysis using only lead I electrocardiogram as an input to assess the feasibility of remote monitoring via wearables that can capture single-lead electrocardiogram data. RESULTS: The University of Tennessee Health Science Center heart failure cohort comprised 346,339 electrocardiograms from 142,601 patients. In this cohort, 60% of participants were Black and 37% were White, with an average age (standard deviation) of 53 (19) years. The heart failure detection model achieved an area under the curve of 0.92 on the holdout set. We then tested the ability of the heart failure model to detect peripartum cardiomyopathy in an independent University of Tennessee Health Science Center cohort of pregnant women and an external Atrium Health Wake Forest Baptist cohort of pregnant women. The independent University of Tennessee Health Science Center cohort included 158 electrocardiograms from 115 patients; our deep-learning model achieved an area under the curve of 0.83 (0.77-0.89) for this data set. The external Atrium Health Wake Forest Baptist cohort involved 80 electrocardiograms from 43 patients; our deep-learning model achieved an area under the curve of 0.94 (0.91-0.98) for this data set. For identifying peripartum cardiomyopathy diagnosed ≥10 days after delivery, the model achieved an area under the curve of 0.88 (0.81-0.94) for the University of Tennessee Health Science Center cohort and of 0.96 (0.93-0.99) for the Atrium Health Wake Forest Baptist cohort. When we repeated our analysis by building a heart failure detection model using only lead-I electrocardiograms, we obtained similarly high detection accuracies, with areas under the curve of 0.73 and 0.93 for the University of Tennessee Health Science Center and Atrium Health Wake Forest Baptist cohorts, respectively. CONCLUSION: Artificial intelligence can accurately detect peripartum cardiomyopathy from electrocardiograms alone. A simple electrocardiographic artificial intelligence-based peripartum screening could result in a timelier diagnosis. Given that results with 1-lead electrocardiogram data were similar to those obtained using all 12 leads, future studies will focus on remote screening for peripartum cardiomyopathy using smartwatches that can capture single-lead electrocardiogram data.
Asunto(s)
Inteligencia Artificial , Cardiomiopatías , Aprendizaje Profundo , Electrocardiografía , Insuficiencia Cardíaca , Periodo Periparto , Complicaciones Cardiovasculares del Embarazo , Humanos , Femenino , Embarazo , Electrocardiografía/métodos , Adulto , Cardiomiopatías/diagnóstico , Cardiomiopatías/fisiopatología , Estudios Retrospectivos , Persona de Mediana Edad , Insuficiencia Cardíaca/diagnóstico , Insuficiencia Cardíaca/fisiopatología , Insuficiencia Cardíaca/epidemiología , Complicaciones Cardiovasculares del Embarazo/diagnóstico , Complicaciones Cardiovasculares del Embarazo/fisiopatología , Curva ROCRESUMEN
BACKGROUND: The use of traditional models to predict heart failure (HF) has limitations in preventing HF hospitalizations. Artificial intelligence (AI) and machine learning (ML) in cardiovascular medicine only have limited data published regarding HF populations, with none assessing the favorability of decongestive therapy aquapheresis (AQ). AI and ML can be leveraged to design non-traditional models to identify those who are at high risk of HF readmissions. OBJECTIVES: This study aimed to develop a model for pretreatment identification of risk for 90-day HF events among HF patients who have undergone AQ. METHODS: Using data from the AVOID-HF (Aquapheresis versus Intravenous Diuretics and Hospitalization for Heart Failure) trial, we designed a ML-based predictive model that can be used before initiating AQ to anticipate who will respond well to AQ and who will be at high risk of future HF events. RESULTS: Using ML we identified the top ten predictors for 90-day HF events. Interestingly, the variable for 'intimate relationships with loved ones' strongly predicted response to therapy. This ML-model was more successful in predicting the outcome in HF patients who were treated with AQ. In the original AVOID-HF trial, the overall 90-day HF event rate in the AQ arm was 32%. Our proposed predictive model was accurate in anticipating 90-day HF events with better statistical accuracy (area under curve 0.88, sensitivity 80%, specificity 75%, negative predictive value 90%, and positive predictive value 57%). CONCLUSIONS: ML can help identify HF patients who will respond to AQ therapy. Our model can identify super-respondents to AQ therapy and predict 90-day HF events better than currently existing traditional models. CONDENSED ABSTRACT: Utilizing data from the AVOID-HF trial, we designed a ML-predictive model that can be used before initiating AQ to anticipate who will respond well to AQ and who will be at high risk of future HF events. Using ML, we identified the top 10 predictors for 90-day HF events. Our model can identify super-respondents to ultrafiltration therapy and predict 90-day HF events better than currently existing traditional models.
Asunto(s)
Insuficiencia Cardíaca , Ultrafiltración , Humanos , Inteligencia Artificial , Insuficiencia Cardíaca/diagnóstico , Insuficiencia Cardíaca/terapia , Hospitalización , Readmisión del PacienteRESUMEN
Background: Cardiovascular diseases contribute to premature mortality globally, resulting in substantial social and economic burdens. The Global Burden of Disease (GBD) Study reported that in 2019 alone, heart attack and strokes accounted for the deaths of 18.6 million individuals. Ischemic heart diseases, including acute myocardial infarction (AMI), accounted for 182 million disability-adjusted life years (DALYs) and it is leading cause of death worldwide. Aim: The aim of this study is to present the burden of AMI in Kazakhstan and describe the outcome of hospitalized patients. Methods: The data of 79,172 people admitted to hospital with ICD-10 diagnosis I21 between 2014 and 2019 was derived from the Unified National Electronic Health System and retrospectively analyzed. Results: The majority of the cohort (53,285, 67%) were men, with an average age of 63 (±12) years, predominantly of Kazakh (38,057, 48%) and Russian (24,583, 31%) ethnicities. Hypertension was the most common comorbidity (61,972, 78%). In males, a sharp increase in incidence is present after 40 years, while for females, the morbidity increases gradually after 55. Throughout the observation period, all-cause mortality rose from 101 to 210 people per million population (PMP). In 2019, AMI account for 169,862 DALYs in Kazakhstan, with a significant proportion (79%) attributed to years of life lost due to premature death (YLDs). Approximately half of disease burden due to AMI (80,794 DALYs) was in age group 55-69 years. Although incidence is higher for men, they have better survival rates than women. In terms of revascularization procedures, coronary artery bypass grafting yielded higher survival rates compared to percutaneous coronary intervention (86.3% and 80.9% respectively) during the 5-year follow-up. Conclusion: This research evaluated the burden and disability-adjusted life years of AMI in Kazakhstan, the largest Central Asian country. The results show that more effective disease management systems and preventive measures at earlier ages are needed.
RESUMEN
RATIONALE: Bronchopulmonary dysplasia (BPD) is the most common morbidity affecting very preterm infants. Gut fungal and bacterial microbial communities contribute to multiple lung diseases and may influence BPD pathogenesis. METHODS: We performed a prospective, observational cohort study comparing the multikingdom fecal microbiota of 144 preterm infants with or without moderate to severe BPD by sequencing the bacterial 16S and fungal ITS2 ribosomal RNA gene. To address the potential causative relationship between gut dysbiosis and BPD, we used fecal microbiota transplant in an antibiotic-pseudohumanized mouse model. Comparisons were made using RNA sequencing, confocal microscopy, lung morphometry, and oscillometry. RESULTS: We analyzed 102 fecal microbiome samples collected during the second week of life. Infants who later developed BPD showed an obvious fungal dysbiosis as compared to infants without BPD (NoBPD, p = 0.0398, permutational multivariate ANOVA). Instead of fungal communities dominated by Candida and Saccharomyces, the microbiota of infants who developed BPD were characterized by a greater diversity of rarer fungi in less interconnected community architectures. On successful colonization, the gut microbiota from infants with BPD augmented lung injury in the offspring of recipient animals. We identified alterations in the murine intestinal microbiome and transcriptome associated with augmented lung injury. CONCLUSIONS: The gut fungal microbiome of infants who will develop BPD is dysbiotic and may contribute to disease pathogenesis.
RESUMEN
Little is known about electrocardiogram (ECG) markers of Parkinson's disease (PD) during the prodromal stage. The aim of the study was to build a generalizable ECG-based fully automatic artificial intelligence (AI) model to predict PD risk during the prodromal stage, up to 5 years before disease diagnosis. This case-control study included samples from Loyola University Chicago (LUC) and University of Tennessee-Methodist Le Bonheur Healthcare (MLH). Cases and controls were matched according to specific characteristics (date, age, sex and race). Clinical data were available from May, 2014 onward at LUC and from January, 2015 onward at MLH, while the ECG data were available as early as 1990 in both institutes. PD was denoted by at least two primary diagnostic codes (ICD9 332.0; ICD10 G20) at least 30 days apart. PD incidence date was defined as the earliest of first PD diagnostic code or PD-related medication prescription. ECGs obtained at least 6 months before PD incidence date were modeled to predict a subsequent diagnosis of PD within three time windows: 6 months-1 year, 6 months-3 years, and 6 months-5 years. We applied a novel deep neural network using standard 10-s 12-lead ECGs to predict PD risk at the prodromal phase. This model was compared to multiple feature engineering-based models. Subgroup analyses for sex, race and age were also performed. Our primary prediction model was a one-dimensional convolutional neural network (1D-CNN) that was built using 131 cases and 1058 controls from MLH, and externally validated on 29 cases and 165 controls from LUC. The model was trained on 90% of the MLH data, internally validated on the remaining 10% and externally validated on LUC data. The best performing model resulted in an external validation AUC of 0.67 when predicting future PD at any time between 6 months and 5 years after the ECG. Accuracy increased when restricted to ECGs obtained within 6 months to 3 years before PD diagnosis (AUC 0.69) and was highest when predicting future PD within 6 months to 1 year (AUC 0.74). The 1D-CNN model based on raw ECG data outperformed multiple models built using more standard ECG feature engineering approaches. These results demonstrate that a predictive model developed in one cohort using only raw 10-s ECGs can effectively classify individuals with prodromal PD in an independent cohort, particularly closer to disease diagnosis. Standard ECGs may help identify individuals with prodromal PD for cost-effective population-level early detection and inclusion in disease-modifying therapeutic trials.
Asunto(s)
Aprendizaje Profundo , Enfermedad de Parkinson , Humanos , Inteligencia Artificial , Estudios de Casos y Controles , Enfermedad de Parkinson/diagnóstico , Síntomas Prodrómicos , ElectrocardiografíaRESUMEN
Background: Heart failure (HF) is a progressive condition with high global incidence. HF has two main subtypes: HF with preserved ejection fraction (HFpEF) and HF with reduced ejection fraction (HFrEF). There is an inherent need for simple yet effective electrocardiogram (ECG)-based artificial intelligence (AI; ECG-AI) models that can predict HF risk early to allow for risk modification. Objective: The main objectives were to validate HF risk prediction models using Multi-Ethnic Study of Atherosclerosis (MESA) data and assess performance on HFpEF and HFrEF classification. Methods: There were six models in comparision derived using ARIC data. 1) The ECG-AI model predicting HF risk was developed using raw 12-lead ECGs with a convolutional neural network. The clinical models from 2) ARIC (ARIC-HF) and 3) Framingham Heart Study (FHS-HF) used 9 and 8 variables, respectively. 4) Cox proportional hazards (CPH) model developed using the clinical risk factors in ARIC-HF or FHS-HF. 5) CPH model using the outcome of ECG-AI and the clinical risk factors used in CPH model (ECG-AI-Cox) and 6) A Light Gradient Boosting Machine model using 288 ECG Characteristics (ECG-Chars). All the models were validated on MESA. The performances of these models were evaluated using the area under the receiver operating characteristic curve (AUC) and compared using the DeLong test. Results: ECG-AI, ECG-Chars, and ECG-AI-Cox resulted in validation AUCs of 0.77, 0.73, and 0.84, respectively. ARIC-HF and FHS-HF yielded AUCs of 0.76 and 0.74, respectively, and CPH resulted in AUC = 0.78. ECG-AI-Cox outperformed all other models. ECG-AI-Cox provided an AUC of 0.85 for HFrEF and 0.83 for HFpEF. Conclusion: ECG-AI using ECGs provides better-validated predictions when compared to HF risk calculators, and the ECG feature model and also works well with HFpEF and HFrEF classification.
RESUMEN
BACKGROUND: Barth syndrome (BTHS) is a rare X-linked genetic disease that affects multiple systems and leads to complex clinical manifestations. Although a considerable amount of research has focused on the physical aspects of the disease, less has focused on the psychosocial impact and quality of life (QoL) in BTHS. METHODS: The current study investigated caregiver- (n = 10) and self-reported (n = 16) psychological well-being and QoL in a cohort of BTHS-affected patients and families. Participants completed the depression and anxiety components of the Patient-Reported Outcomes Information System (PROMIS) Short Form 8A and Health-related quality of life (HRQoL) surveys at enrollment and again during a follow-up period ranging from 6 to 36 months after baseline. RESULTS: Quality of life changed significantly over time and the various domains with some improvement and some decline. Among the available caregiver-patient dyad data, there was a trend toward discordance between caregiver and self-reported outcomes. Most notably, patients reported improvement in HRQoL, while caregivers reported declines. This suggests that there may be differences in perceived quality of life between the patients and parents, though our study is limited by small sample size. CONCLUSION: Our study provides valuable insights into the impacts of psychosocial and mental health aspects of BTHS. Implications of these findings include incorporating longitudinal assessment of QoL and screening for psychological symptoms in BTHS care to identify interventions that may drastically impact health status and the course of the disease.
RESUMEN
Cardiovascular disease is a leading cause of death among cancer survivors, second only to cancer recurrence or development of new tumors. Cardio-oncology has therefore emerged as a relatively new specialty focused on prevention and management of cardiovascular consequences of cancer therapies. Yet challenges remain regarding precision and accuracy with predicting individuals at highest risk for cardiotoxicity. Barriers such as access to care also limit screening and early diagnosis to improve prognosis. Thus, developing innovative approaches for prediction and early detection of cardiovascular illness in this population is critical. In this review, we provide an overview of the present state of machine learning applications in cardio-oncology. We begin by outlining some factors that should be considered while utilizing machine learning algorithms. We then examine research in which machine learning has been applied to improve prediction of cardiac dysfunction in cancer survivors. We also highlight the use of artificial intelligence (AI) in conjunction with electrocardiogram (ECG) to predict cardiac malfunction and also atrial fibrillation (AF), and we discuss the potential role of wearables. Additionally, the article summarizes future prospects and critical takeaways for the application of machine learning in cardio-oncology. This study is the first in a series on artificial intelligence in cardio-oncology, and complements our manuscript on echocardiography and other forms of imaging relevant to cancer survivors cared for in cardiology clinical practice.
RESUMEN
BACKGROUND: Parkinson's disease (PD) is a chronic, disabling neurodegenerative disorder. OBJECTIVE: To predict a future diagnosis of PD using questionnaires and simple non-invasive clinical tests. METHODS: Participants in the prospective Kuakini Honolulu-Asia Aging Study (HAAS) were evaluated biannually between 1995-2017 by PD experts using standard diagnostic criteria. Autopsies were sought on all deaths. We input simple clinical and risk factor variables into an ensemble-tree based machine learning algorithm and derived models to predict the probability of developing PD. We also investigated relationships of predictive models and neuropathologic features such as nigral neuron density. RESULTS: The study sample included 292 subjects, 25 of whom developed PD within 3 years and 41 by 5 years. 116 (46%) of 251 subjects not diagnosed with PD underwent autopsy. Light Gradient Boosting Machine modeling of 12 predictors correctly classified a high proportion of individuals who developed PD within 3 years (area under the curve (AUC) 0.82, 95%CI 0.76-0.89) or 5 years (AUC 0.77, 95%CI 0.71-0.84). A large proportion of controls who were misclassified as PD had Lewy pathology at autopsy, including 79%of those who died within 3 years. PD probability estimates correlated inversely with nigral neuron density and were strongest in autopsies conducted within 3 years of index date (râ=â-0.57, pâ<â0.01). CONCLUSION: Machine learning can identify persons likely to develop PD during the prodromal period using questionnaires and simple non-invasive tests. Correlation with neuropathology suggests that true model accuracy may be considerably higher than estimates based solely on clinical diagnosis.
Asunto(s)
Enfermedad de Parkinson , Humanos , Aprendizaje Automático , Enfermedad de Parkinson/diagnóstico , Enfermedad de Parkinson/patología , Síntomas Prodrómicos , Estudios Prospectivos , Factores de RiesgoRESUMEN
OBJECTIVE: To understand the temporal relationships of postoperative complications in children and determine if they are related to each other in a predictable manner. SUMMARY OF BACKGROUND DATA: Children with multiple postoperative complications have increased suffering and higher risk for mortality. Rigorous analysis of the temporal relations between complications, how complications might cluster, and the implications of such clusters for children have not been published. Herein, we analyze the relationships between postoperative complications in children. METHODS: Data source: Surgical operations included in the National Surgical Quality Improvement Program Pediatric Participant Use Data File from 2013 to 2017. The main outcomes measure was presence of 1 or more postoperative complications within 30âdays of surgery. Operations followed by multiple complications were analyzed using network analysis to study prevalence, timing, and co-occurrences of clusters of complications. RESULTS: This study cohort consisted of 432,090 operations; 388,738 (89.97%) had no postoperative complications identified, 36,105 (8.35%) operations resulted in 1 postoperative complication and 7247 (1.68%) operations resulted in 2 or more complications. Patients with multiple complications were more likely to be younger, male, African American, with a higher American Society of Anesthesiologists score, and to undergo nonelective operations (P < 0.001). More patients died with 2 complication versus 1 complication vs no complication (5.3% vs 1.5% vs 0.14%, P < 0.001). Network analysis identified 4 Louvain clusters of complications with dense intracluster relationships. CONCLUSIONS: Children with multiple postoperative complications are at higher risk of death, than patients with no complication, or a single complication. Multiple complications are grouped into defined clusters and are not independent.
Asunto(s)
Complicaciones Posoperatorias , Mejoramiento de la Calidad , Niño , Estudios de Cohortes , Humanos , Masculino , Complicaciones Posoperatorias/epidemiología , Complicaciones Posoperatorias/etiología , Periodo Posoperatorio , Estudios Retrospectivos , Factores de RiesgoRESUMEN
Study objective: To determine whether there has been growth in publications on the use of artificial intelligence in cardiology and oncology, we assessed historical trends in publications related to artificial intelligence applications in cardiology and oncology, which are the two fields studying the leading causes of death worldwide. Upward trends in publications may indicate increasing interest in the use of artificial intelligence in these crucial fields. Design/setting: To evaluate evidence of increasing publications on the use of artificial intelligence in cardiology and oncology, historical trends in related publications on PubMed (the biomedical repository most frequently used by clinicians and scientists in these fields) were reviewed. Results: Findings indicated that research output related to artificial intelligence (and its subcategories) generally increased over time, particularly in the last five years. With some initial degree of vacillation in publication trends, a slight qualitative inflection was noted in approximately 2015, in general publications and especially for oncology and cardiology, with subsequent consistent exponential growth. Publications predominantly focused on "machine learning" (n = 20,301), which contributed to the majority of the accelerated growth in the field, compared to "artificial intelligence" (n = 4535), "natural language processing" (n = 2608), and "deep learning" (n = 4459). Conclusion: Trends in the general biomedical literature and particularly in cardiology and oncology indicated exponential growth over time. Further exponential growth is expected in future years, as awareness and cross-disciplinary collaboration and education increase. Publications specifically on machine learning will likely continue to lead the way.
RESUMEN
BACKGROUND: Patients admitted to the emergency department (ED) with COVID-19 symptoms are routinely required to have chest radiographs and computed tomography (CT) scans. COVID-19 infection has been directly related to the development of acute respiratory distress syndrome (ARDS) and severe infections could lead to admission to intensive care and increased risk of death. The use of clinical data in machine learning models available at time of admission to ED can be used to assess possible risk of ARDS, the need for intensive care (admission to the Intensive Care Unit; ICU) as well as risk of mortality. In addition, chest radiographs can be inputted into a deep learning model to further assess these risks. PURPOSE: This research aimed to develop machine and deep learning models using both structured clinical data and image data from the electronic health record (EHR) to predict adverse outcomes following ED admission. MATERIALS AND METHODS: Light Gradient Boosting Machine (LightGBM) was used as the main machine learning algorithm using all clinical data including 42 variables. Compact models were also developed using the 15 most important variables to increase applicability of the models in clinical settings. To predict risk (or early stratified risk) of the aforementioned health outcome events, transfer learning from the CheXNet model was also implemented on the available data. This research utilized clinical data and chest radiographs of 3,571 patients, 18 years and older, admitted to the emergency department between 9th March 2020 and 29th October 2020 at Loyola University Medical Center. MAIN FINDINGS: The research results show that we can detect COVID-19 infection (AUC = 0.790 (0.746-0.835)), predict the risk of developing ARDS (AUC = 0.781 (0.690-0.872), risk stratification of the need for ICU admission (AUC = 0.675 (0.620-0.713)) and mortality (AUC = 0.759 (0.678-0.840)) at moderate accuracy from both chest X-ray images and clinical data. PRINCIPAL CONCLUSIONS: The results can help in clinical decision making, especially when addressing ARDS and mortality, during the assessment of patients admitted to the ED with or without COVID-19 symptoms.
RESUMEN
BACKGROUND: Patients with advanced chronic kidney disease (CKD) are at high risk for dyskalemias, which may induce arrhythmias that require immediate emergent or hospital care. The association of dyskalemias with short-term hospital/emergency room (ER) visits in advanced CKD is understudied. OBJECTIVE: To assess the association of dyskalemias with short-term hospital/ER visits in an advanced CKD population. METHODS: From among 102,477 US veterans transitioning to dialysis from 2007 to 2015, we identified 21,366 patients with 2 predialysis outpatient eGFR < 30 ml/min/1.73m2 90-365 days apart (with the second eGFR serving as the index date) and at least 1 potassium (K) in the baseline period (1 year before index) and 1 outpatient K (oK) in the follow-up (1 year after the index but before dialysis initiation). We examined the association of time-varying hypokalemia (K < 3.5 mEq/L) and hyperkalemia (K > 5.5 mEq/L) vs referent (3.5-5.5 mEq/L) with separate hospital and ER visits within 2 calendar days following each oK value over the 1-year follow-up period from the index. We used generalized estimating equations with binary distribution and logit link to model the exposure-outcome relationship adjusted for various confounders. We conducted various subgroup and sensitivity analyses to test the robustness of our results. RESULTS: Over the 1-year follow-up, 125,266 oK measurements were observed, of which 6.8% and 3.7% were classified as hyper- and hypokalemia, respectively. In the multivariable-adjusted model, hyperkalemia (adjusted odds ratio [aOR] = 2.04; 95% CI = 1.88-2.21) and hypokalemia (aOR = 1.66; 95% CI = 1.48-1.86) were associated with significantly higher odds of hospital visits. Similarly, hyperkalemia (aOR = 1.83; 95% CI = 1.65-2.03) and hypokalemia (aOR = 1.24; 95% CI = 1.07-1.44) were associated with significantly higher odds of ER visits. Results were robust to subgroups and sensitivity analyses. CONCLUSIONS: In patients with advanced CKD, dyskalemias are associated with higher risk of hospital/ER visits. Interventions targeted at lowering the risk of dyskalemias might help in reducing the health care utilization and associated economic burden among patients with advanced CKD experiencing dyskalemias. DISCLOSURES: This study was supported by grant 5U01DK102163 from the National Institute of Health (NIH) to Kamyar Kalantar-Zadeh and Csaba P. Kovesdy and by resources from the US Department of Veterans Affairs. The data reported here have been supplied in part by the United States Renal Data System (USRDS). Support for VA/CMS data were provided by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Health Services Research and Development, VA Information Resource Center (project numbers SDR 02-237 and 98-004). Opinions expressed in this article are those of the authors and do not necessarily represent the opinion of the Department of Veterans Affairs or the funding institution. Kovesdy has received honoraria from Akebia, Ardelyx, Astra Zeneca, Bayer, Boehringer-Ingelheim, Cara Therapeutics, Reata, and Tricida unrelated to this study. Kalantar-Zadeh has received honoraria and/or support from Abbott, Abbvie, ACI Clinical (Cara Therapeutics), Akebia, Alexion, Amgen, American Society of Nephrology, Astra-Zeneca, Aveo, BBraun, Chugai, Cytokinetics, Daiichi, DaVita, Fresenius, Genentech, Haymarket Media, Hofstra Medical School, International Federation of Kidney Foundations, International Society of Hemodialysis, International Society of Renal Nutrition & Metabolism, Japanese Society of Dialysis Therapy, Hospira, Kabi, Keryx, Kissei, Novartis, OPKO, National Institutes of Health, National Kidney Foundations, Pfizer, Regulus, Relypsa, Resverlogix, Dr Schaer, Sandoz, Sanofi, Shire, Veterans Affairs, Vifor, UpToDate, and ZS-Pharma, unrelated to this study. Gatwood has received research support from AstraZeneca, Merck & Co., and GlaxoSmithKline unrelated to this study. Obi has received research support from Relypsa/Vifor Pharma Inc. The remaining authors declare that they have no relevant financial interests.
Asunto(s)
Costos de la Atención en Salud , Hiperpotasemia/fisiopatología , Aceptación de la Atención de Salud , Insuficiencia Renal Crónica/patología , Anciano , Bases de Datos Factuales , Femenino , Humanos , Masculino , Persona de Mediana Edad , Estudios Retrospectivos , Estados UnidosRESUMEN
BACKGROUND AND PURPOSE: Early identification of large vessel occlusions (LVO) and timely recanalization are paramount to improved clinical outcomes in acute ischemic stroke. A stroke assessment that maximizes sensitivity and specificity for LVOs is needed to identify these cases and not overburden the health system with unnecessary transfers. Machine learning techniques are being used for predictive modeling in many aspects of stroke care and may have potential in predicting LVO presence and mechanical thrombectomy (MT) candidacy. METHODS: Ischemic stroke patients treated at Loyola University Medical Center from July 2018 to June 2019 (N = 286) were included. Thirty-five clinical and demographic variables were analyzed using machine learning algorithms, including logistic regression, extreme gradient boosting, random forest (RF), and decision trees to build models predictive of LVO presence and MT candidacy by area of the curve (AUC) analysis. The best performing model was compared with prior stroke scales. RESULTS: When using all 35 variables, RF best predicted LVO presence (AUC = 0.907 ± 0.856-0.957) while logistic regression best predicted MT candidacy (AUC = 0.930 ± 0.886-0.974). When compact models were evaluated, a 10-feature RF model best predicted LVO (AUC = 0.841 ± 0.778-0.904) and an 8-feature RF model best predicted MT candidacy (AUC = 0.862 ± 0.782-0.942). The compact RF models had sensitivity, specificity, negative predictive value and positive predictive value of 0.81, 0.87, 0.92, 0.72 for LVO and 0.87, 0.97, 0.97, 0.86 for MT, respectively. The 10-feature RF model was superior at predicting LVO to all previous stroke scales (AUC 0.944 vs 0.759-0.878) and the 8-feature RF model was superior at predicting MT (AUC 0.970 vs 0.746-0.834). CONCLUSION: Random forest machine learning models utilizing clinical and demographic variables predicts LVO presence and MT candidacy with a high degree of accuracy in an ischemic stroke cohort. Further validation of this strategy for triage of stroke patients requires prospective and external validation.
Asunto(s)
Isquemia Encefálica , Accidente Cerebrovascular Isquémico , Accidente Cerebrovascular , Isquemia Encefálica/diagnóstico , Isquemia Encefálica/cirugía , Humanos , Aprendizaje Automático , Masculino , Estudios Prospectivos , Estudios Retrospectivos , Accidente Cerebrovascular/diagnóstico , Accidente Cerebrovascular/cirugía , Trombectomía , Activador de Tejido PlasminógenoRESUMEN
INTRODUCTION: Hypo- and hyperkalemia are associated with a higher risk of ischemic stroke. However, this association has not been examined in an advanced chronic kidney disease (CKD) population. METHODS: From among 102,477 US veterans transitioning to dialysis between 2007 and 2015, 21,357 patients with 2 pre-dialysis outpatient estimated glomerular filtration rates <30 mL/min/1.73 m2 90-365 days apart and at least 1 potassium (K) each in the baseline and follow-up period were identified. We separately examined the association of both baseline time-averaged K (chronic exposure) and time-updated K (acute exposure) treated as categorized (hypokalemia [K <3.5 mEq/L] and hyperkalemia [K >5.5 mEq/L] vs. referent [3.5-5.5 mEq/L]) and continuous exposure with time to the first ischemic stroke event prior to dialysis initiation using multivariable-adjusted Cox regression models. RESULTS: A total of 2,638 (12.4%) ischemic stroke events (crude event rate 41.9 per 1,000 patient years; 95% confidence interval [CI] 40.4-43.6) over a median (Q1-Q3) follow-up time of 2.56 (1.59-3.89) years were observed. The baseline time-averaged K category of hypokalemia (adjusted hazard ratio [aHR], 95% CI: 1.35, 1.01-1.81) was marginally associated with a significantly higher risk of ischemic stroke. However, time-updated hyperkalemia was associated with a significantly lower risk of ischemic stroke (aHR, 95% CI: 0.82, 0.68-0.98). The exposure-outcome relationship remained consistent when using continuous K levels for both the exposures. DISCUSSION/CONCLUSION: In patients with advanced CKD, hypokalemia (chronic exposure) was associated with a higher risk of ischemic stroke, whereas hyperkalemia (acute exposure) was associated with a lower risk of ischemic stroke. Further studies in this population are needed to explore the mechanisms underlying these associations.