RESUMEN

OBJECTIVES: Enhancing critical care efficacy involves evaluating and improving system functioning. Benchmarking, a retrospective comparison of results against standards, aids risk-adjusted assessment and helps healthcare providers identify areas for improvement based on observed and predicted outcomes. The last two decades have seen the development of several models using machine learning (ML) for clinical outcome prediction. ML is a field of artificial intelligence focused on creating algorithms that enable computers to learn from and make predictions or decisions based on data. This narrative review centers on key discoveries and outcomes to aid clinicians and researchers in selecting the optimal methodology for critical care benchmarking using ML. METHODS: We used PubMed to search the literature from 2003 to 2023 regarding predictive models utilizing ML for mortality (592 articles), length of stay (143 articles), or mechanical ventilation (195 articles). We supplemented the PubMed search with Google Scholar, making sure relevant articles were included. Given the narrative style, papers in the cohort were manually curated for a comprehensive reader perspective. RESULTS: Our report presents comparative results for benchmarked outcomes and emphasizes advancements in feature types, preprocessing, model selection, and validation. It showcases instances where ML effectively tackled critical care outcome-prediction challenges, including nonlinear relationships, class imbalances, missing data, and documentation variability, leading to enhanced results. CONCLUSIONS: Although ML has provided novel tools to improve the benchmarking of critical care outcomes, areas that require further research include class imbalance, fairness, improved calibration, generalizability, and long-term validation of published models.

RESUMEN

OBJECTIVES: Electronic health records enable automated data capture for risk models but may introduce bias. We present the Philips Critical Care Outcome Prediction Model (CCOPM) focused on addressing model features sensitive to data drift to improve benchmarking ICUs on mortality performance. DESIGN: Retrospective, multicenter study of ICU patients randomized in 3:2 fashion into development and validation cohorts. Generalized additive models (GAM) with features designed to mitigate biases introduced from documentation of admission diagnosis, Glasgow Coma Scale (GCS), and extreme vital signs were developed using clinical features representing the first 24 hours of ICU admission. SETTING: eICU Research Institute database derived from ICUs participating in the Philips eICU telecritical care program. PATIENTS: A total of 572,985 adult ICU stays discharged from the hospital between January 1, 2017, and December 31, 2018, were included, yielding 509,586 stays in the final cohort; 305,590 and 203,996 in development and validation cohorts, respectively. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Model discrimination was compared against Acute Physiology and Chronic Health Evaluation (APACHE) IVa/IVb models on the validation cohort using the area under the receiver operating characteristic (AUROC) curve. Calibration assessed by actual/predicted ratios, calibration-in-the-large statistics, and visual analysis. Performance metrics were further stratified by subgroups of admission diagnosis and ICU characteristics. Historic data from two health systems with abrupt changes in Glasgow Coma Scale (GCS) documentation were assessed in the year prior to and after data shift. CCOPM outperformed APACHE IVa/IVb for ICU mortality (AUROC, 0.925 vs 0.88) and hospital mortality (AUROC, 0.90 vs 0.86). Better calibration performance was also attained among subgroups of different admission diagnoses, ICU types, and over unique ICU-years. The CCOPM provided more stable predictions compared with APACHE IVa within an external cohort of greater than 120,000 patients from two health systems with known changes in GCS documentation. CONCLUSIONS: These mortality risk models demonstrated excellent performance compared with APACHE while appearing to mitigate bias introduced through major shifts in GCS documentation at two large health systems. This provides evidence to support using automated capture rather than trained personnel for capture of GCS data used in benchmarking ICUs on mortality performance.

Asunto(s)

RESUMEN

OBJECTIVES: Evaluate the accuracy of different ICU risk models repurposed as continuous markers of severity of illness. DESIGN: Nonintervention cohort study. SETTING: eICU Research Institute ICUs using tele-ICU software calculating continuous ICU Discharge Readiness Scores between January 2013 and March 2016. PATIENTS: Five hundred sixty-one thousand four hundred seventy-eight adult ICU patients with an ICU length of stay between 4 hours and 30 days. INTERVENTIONS: Not available. MEASUREMENTS AND MAIN RESULTS: Hourly Acute Physiology and Chronic Health Evaluation IV, Sequential Organ Failure Assessment, and Discharge Readiness Scores were calculated beginning hour 4 of the ICU stay. Primary outcome was the area under the receiver operating characteristic curve for the mean score with ICU mortality. Secondary outcomes included area under the receiver operating characteristic curves for ICU mortality with admission, median, maximum and last scores, and for death within 24 hours. The trajectories of each score were visualized by plotting the hourly averages against time in the ICU, stratified by mortality and length of stay. The area under the receiver operating characteristic curves for mean Acute Physiology and Chronic Health Evaluation, Sequential Organ Failure Assessment, and Discharge Readiness Scores were 0.90 (0.89-0.90), 0.86 (0.86-0.86), and 0.94 (0.94-0.94), respectively. The area under the receiver operating characteristic curves for hourly Acute Physiology and Chronic Health Evaluation, Sequential Organ Failure Assessment, and Discharge Readiness Scores predicting 24-hour mortality were 0.81 (0.81-0.81), 0.76 (0.76-0.76), and 0.86 (0.86-0.86). Discharge Readiness Scores had a higher area under the receiver operating characteristic curve than both Acute Physiology and Chronic Health Evaluation and Sequential Organ Failure Assessment for each metric. Acute Physiology and Chronic Health Evaluation and Sequential Organ Failure Assessment scores increased throughout the first 24 hours in both survivors and nonsurvivors; Discharge Readiness Scores continuously decreased in survivors and temporarily decreased before increasing by hour 36 in nonsurvivors with longer length of stays. CONCLUSIONS: Acute Physiology and Chronic Health Evaluation, Sequential Organ Failure Assessment, and Discharge Readiness Scores all have relatively high discrimination for ICU mortality when used continuously; Discharge Readiness Scores tended to have slightly higher area under the receiver operating characteristic curves for each endpoint. These findings validate the use of these models on a population level for continuous risk adjustment in the ICU, although Acute Physiology and Chronic Health Evaluation and Sequential Organ Failure Assessment appear slower to respond to improvements in patient status than Discharge Readiness Scores, and Discharge Readiness Scores may reflect physiologic improvement from interventions, potentially underestimating risk.

Asunto(s)

RESUMEN

Dyspnea is a nonspecific symptom of any disease involving the respiratory system. Although diseases of the lungs, chest wall, pleura, diaphragm, upper airway, and heart are most common, diseases of many other organ systems (eg, neuromuscular, skeletal, renal, endocrine, rheumatologic, hematologic, and psychiatric) may involve the respiratory system and present with dyspnea. Dyspnea should be evaluated systematically, and a thorough history and physical examination and baseline tests of heart and lung function are necessary to establish a complete database. More sophisticated testing may be needed when the cause is not readily apparent from the initial work-up. Treatment is best and most effective when geared toward a specific etiology, but if this is not possible, nonspecific treatment of the symptom pf dyspnea may afford the patient some benefit.

Asunto(s)

RESUMEN

The glucocorticoid receptor (GR) gene (NR3C1) maps to 5q31, a region genetically linked to asthma. In this study, NR3C1 exons 1A, 1B, and exons 1C to 9 (alpha and beta) were sequenced in a screening panel of asthmatics and unaffected controls from US Caucasian, African American, US Hispanic, and Dutch Caucasian populations to identify polymorphisms for genetic association studies. Eight polymorphisms were identified in exon 1A, but none were located in putative transcription regulatory sites. Thirty-four polymorphisms were identified in exons 1B to 9 (alpha and beta), 17 of which were novel. Eight coding polymorphisms were identified (4 non-synonymous). One novel mutation (Ala229Thr) was identified in a Hispanic individual. Linkage disequilibrium (LD) was strongest between polymorphisms spanning intron 2 to exon 9beta. This data shows the variability of NR3C1 polymorphism frequencies between racial groups and confirms that NR3C1 non-synonymous coding polymorphisms are generally rare in mild/moderate asthmatics and unaffected controls.

Asunto(s)