ABSTRACT
OBJECTIVES: To develop proof-of-concept algorithms using alternative approaches to capture provider sentiment in ICU notes. DESIGN: Retrospective observational cohort study. SETTING: The Multiparameter Intelligent Monitoring of Intensive Care III (MIMIC-III) and the University of California, San Francisco (UCSF) deidentified notes databases. PATIENTS: Adult (≥18 yr old) patients admitted to the ICU. MEASUREMENTS AND MAIN RESULTS: We developed two sentiment models: 1) a keywords-based approach using a consensus-based clinical sentiment lexicon comprised of 72 positive and 103 negative phrases, including negations and 2) a Decoding-enhanced Bidirectional Encoder Representations from Transformers with disentangled attention-v3-based deep learning model (keywords-independent) trained on clinical sentiment labels. We applied the models to 198,944 notes across 52,997 ICU admissions in the MIMIC-III database. Analyses were replicated on an external sample of patients admitted to a UCSF ICU from 2018 to 2019. We also labeled sentiment in 1,493 note fragments and compared the predictive accuracy of our tools to three popular sentiment classifiers. Clinical sentiment terms were found in 99% of patient visits across 88% of notes. Our two sentiment tools were substantially more predictive (Spearman correlations of 0.62-0.84, p values < 0.00001) of labeled sentiment compared with general language algorithms (0.28-0.46). CONCLUSION: Our exploratory healthcare-specific sentiment models can more accurately detect positivity and negativity in clinical notes compared with general sentiment tools not designed for clinical usage.
ABSTRACT
Study Objectives: Following extubation in the intensive care unit (ICU), upper airway (UA) edema and respiratory depressants may promote UA dysfunction. We tested the hypothesis that opioids increase the risk of sleep apnea early after extubation. Methods: Fifty-six ICU patients underwent polysomnography the night after extubation. Airflow limitation during wakefulness was identified using bedside spirometry. Correlation and ordinal regression analyses were used to quantify the effects of preextubation opioid dose on postextubation apnea-hypopnea index (AHI) and severity of sleep apnea and whether or not inspiratory airway obstruction (ratio of maximum expiratory and inspiratory airflows at 50% of vital capacity [MEF50/MIF50] ≥ 1) during wakefulness predicts airway obstruction during sleep. Data were adjusted for age, gender, body mass index, as well as a generalized propensity score balanced for APACHE II, score for preoperative prediction of obstructive sleep apnea, duration of mechanical ventilation, chronic obstructive pulmonary disease, and a procedural severity score for morbidity. Results: Sleep apnea (AHI ≥ 5) was present in 40 (71%) of the 56 patients. Morphine equivalent dose given 24 hours prior extubation predicted obstructive respiratory events during sleep (r = 0.35, p = .01) and sleep apnea (odds ratio [OR] 1.17; 95% confidence interval [CI] 1.02-1.34). Signs of inspiratory UA obstruction (MEF50/MIF50 ≥ 1) assessed by bedside spirometry were strongly associated with sleep apnea (OR 5.93; 95% CI 1.16-30.33). Conclusions: High opioid dose given 24 hours prior to extubation increases the likelihood of postextubation sleep apnea in the ICU, particularly in patients with anatomical vulnerability following extubation.
Subject(s)
Airway Extubation/adverse effects , Airway Obstruction/chemically induced , Analgesics, Opioid/adverse effects , Respiration, Artificial/adverse effects , Sleep Apnea, Obstructive/chemically induced , Adult , Aged , Analgesics, Opioid/therapeutic use , Body Mass Index , Exhalation , Female , Humans , Intensive Care Units , Lung , Male , Middle Aged , Perioperative Period , Polysomnography , Prospective Studies , Sleep/physiology , Spirometry , Tidal Volume , Wakefulness/physiologyABSTRACT
OBJECTIVES: Inhalational anesthetics are bronchodilators with immunomodulatory effects. We sought to determine the effect of inhalational anesthetic dose on risk of severe postoperative respiratory complications. DESIGN: Prospective analysis of data on file in surgical cases between January 2007 and December 2015. SETTING: Massachusetts General Hospital (tertiary referral center) and two affiliated community hospitals. PATIENTS: A total of 124,497 adult patients (105,267 in the study cohort and 19,230 in the validation cohort) undergoing noncardiac surgical procedures and requiring general anesthesia with endotracheal intubation. INTERVENTIONS: Median effective dose equivalent of inhalational anesthetics during surgery (derived from mean end-tidal inhalational anesthetic concentrations). MEASUREMENTS AND MAIN RESULTS: Postoperative respiratory complications occurred in 6,979 of 124,497 cases (5.61%). High inhalational anesthetic dose of 1.20 (1.13-1.30) (median [interquartile range])-fold median effective dose equivalent versus 0.57 (0.45-0.64)-fold median effective dose equivalent was associated with lower odds of postoperative respiratory complications (odds ratio, 0.59; 95% CI, 0.53-0.65; p < 0.001). Additionally, high inhalational anesthetic dose was associated with lower 30-day mortality and lower cost. Inhalational anesthetic dose increase and reduced risk of postoperative respiratory complications remained significant in sensitivity analyses stratified by preoperative and intraoperative risk factors. CONCLUSIONS: Intraoperative use of higher inhalational anesthetic doses is strongly associated with lower odds of postoperative respiratory complications, lower 30-day mortality, and lower cost of hospital care. The authors speculate based on these data that sedation with inhalational anesthetics outside of the operating room may likewise have protective effects that decrease the risk of respiratory complications in vulnerable patients.
Subject(s)
Anesthetics, Inhalation/administration & dosage , Pneumonia/prevention & control , Postoperative Complications/prevention & control , Pulmonary Edema/prevention & control , Respiratory Insufficiency/prevention & control , Dose-Response Relationship, Drug , Female , Hospital Costs , Hospital Mortality , Humans , Intubation, Intratracheal , Male , Massachusetts/epidemiology , Middle Aged , Postoperative Complications/epidemiology , Propensity Score , Prospective Studies , Pulmonary Edema/epidemiology , Respiratory Insufficiency/epidemiologyABSTRACT
Muscle weakness is common in the surgical intensive care unit (ICU). Low muscle mass at ICU admission is a significant predictor of adverse outcomes. The consequences of ICU-acquired muscle weakness depend on the underlying mechanism. Temporary drug-induced weakness when properly managed may not affect outcome. Severe perioperative acquired weakness that is associated with adverse outcomes (prolonged mechanical ventilation, increases in ICU length of stay, and mortality) occurs with persistent (time frame: days) activation of protein degradation pathways, decreases in the drive to the skeletal muscle, and impaired muscular homeostasis. ICU-acquired muscle weakness can be prevented by early treatment of the underlying disease, goal-directed therapy, restrictive use of immobilizing medications, optimal nutrition, activating ventilatory modes, early rehabilitation, and preventive drug therapy. In this article, the authors review the nosology, epidemiology, diagnosis, and prevention of ICU-acquired weakness in surgical ICU patients.
Subject(s)
Critical Care/methods , Intensive Care Units , Muscle Weakness/diagnosis , Muscle Weakness/prevention & control , Critical Illness , Humans , Iatrogenic Disease , Length of Stay/statistics & numerical data , Muscle Weakness/epidemiology , Respiration, Artificial/statistics & numerical dataABSTRACT
Accuracy plays a vital role in the medical field as it concerns with the life of an individual. Extensive research has been conducted on disease classification and prediction using machine learning techniques. However, there is no agreement on which classifier produces the best results. A specific classifier may be better than others for a specific dataset, but another classifier could perform better for some other dataset. Ensemble of classifiers has been proved to be an effective way to improve classification accuracy. In this research we present an ensemble framework with multi-layer classification using enhanced bagging and optimized weighting. The proposed model called "HM-BagMoov" overcomes the limitations of conventional performance bottlenecks by utilizing an ensemble of seven heterogeneous classifiers. The framework is evaluated on five different heart disease datasets, four breast cancer datasets, two diabetes datasets, two liver disease datasets and one hepatitis dataset obtained from public repositories. The analysis of the results show that ensemble framework achieved the highest accuracy, sensitivity and F-Measure when compared with individual classifiers for all the diseases. In addition to this, the ensemble framework also achieved the highest accuracy when compared with the state of the art techniques. An application named "IntelliHealth" is also developed based on proposed model that may be used by hospitals/doctors for diagnostic advice.
Subject(s)
Computer Communication Networks , Decision Support Systems, Clinical , Diagnosis, Computer-Assisted , Medical Informatics Applications , Bayes Theorem , Humans , Machine LearningABSTRACT
BACKGROUND: Duration of action increases with repeated administration of neuromuscular-blocking agents, and intraoperative use of high doses of neuromuscular-blocking agent may affect respiratory safety. METHODS: In a hospital-based registry study on 48,499 patients who received intermediate-acting neuromuscular-blocking agents, the authors tested the primary hypothesis that neuromuscular-blocking agents are dose dependently associated with the risk of postoperative respiratory complications. In the secondary analysis, the authors evaluated the association between neostigmine dose given for reversal of neuromuscular-blocking agents and respiratory complications. Post hoc, the authors evaluated the effects of appropriate neostigmine reversal (neostigmine ≤ 60 µg/kg after recovery of train-of-four count of 2) on respiratory complications. The authors controlled for patient-, anesthesia-, and surgical complexity-related risk factors. RESULTS: High doses of neuromuscular-blocking agents were associated with an increased risk of postoperative respiratory complications (n = 644) compared with low doses (n = 205) (odds ratio [OR], 1.28; 95% CI, 1.04 to 1.57). Neostigmine was associated with a dose-dependent increase in the risk of postoperative respiratory complications (OR, 1.51; 95% CI, 1.25 to 1.83). Post hoc analysis revealed that appropriate neostigmine reversal eliminated the dose-dependent association between neuromuscular-blocking agents and respiratory complications (for neuromuscular-blocking agent effects with appropriate reversal: OR, 0.98; 95% CI, 0.63 to 1.52). CONCLUSIONS: The use of neuromuscular-blocking agents was dose dependently associated with increased risk of postoperative respiratory complications. Neostigmine reversal was also associated with a dose-dependent increase in the risk of respiratory complications. However, the exploratory data analysis suggests that the proper use of neostigmine guided by neuromuscular transmission monitoring results can help eliminate postoperative respiratory complications associated with the use of neuromuscular-blocking agents.
Subject(s)
Neuromuscular Blocking Agents/administration & dosage , Neuromuscular Blocking Agents/adverse effects , Postoperative Complications/chemically induced , Postoperative Complications/epidemiology , Respiration Disorders/chemically induced , Respiration Disorders/epidemiology , Adrenergic alpha-Agonists/therapeutic use , Adult , Aged , Aged, 80 and over , Cholinesterase Inhibitors/adverse effects , Dose-Response Relationship, Drug , Female , Humans , Male , Middle Aged , Neostigmine/adverse effects , Norepinephrine/therapeutic use , Risk Factors , Treatment Outcome , Young AdultABSTRACT
Conventional clinical decision support systems are based on individual classifiers or simple combination of these classifiers which tend to show moderate performance. This research paper presents a novel classifier ensemble framework based on enhanced bagging approach with multi-objective weighted voting scheme for prediction and analysis of heart disease. The proposed model overcomes the limitations of conventional performance by utilizing an ensemble of five heterogeneous classifiers: Naïve Bayes, linear regression, quadratic discriminant analysis, instance based learner and support vector machines. Five different datasets are used for experimentation, evaluation and validation. The datasets are obtained from publicly available data repositories. Effectiveness of the proposed ensemble is investigated by comparison of results with several classifiers. Prediction results of the proposed ensemble model are assessed by ten fold cross validation and ANOVA statistics. The experimental evaluation shows that the proposed framework deals with all type of attributes and achieved high diagnosis accuracy of 84.16 %, 93.29 % sensitivity, 96.70 % specificity, and 82.15 % f-measure. The f-ratio higher than f-critical and p value less than 0.05 for 95 % confidence interval indicate that the results are extremely statistically significant for most of the datasets.
Subject(s)
Algorithms , Heart Diseases/diagnosis , Analysis of Variance , Computer Systems , Databases as Topic , Decision Support Systems, Clinical , HumansABSTRACT
BACKGROUND: This multi-centre, prospective, randomized, double-blind, placebo-controlled study was designed to test the hypotheses that parecoxib improves patients' postoperative analgesia without increasing surgical blood loss following radical open prostatectomy. METHODS: 105 patients (64 ± 7 years old) were randomized to receive either parecoxib or placebo with concurrent morphine patient controlled analgesia. Cumulative opioid consumption (primary objective) and the overall benefit of analgesia score (OBAS), the modified brief pain inventory short form (m-BPI-sf), the opioid-related symptom distress scale (OR-SDS), and perioperative blood loss (secondary objectives) were assessed. RESULTS: In each group 48 patients received the study medication for 48 hours postoperatively. Parecoxib significantly reduced cumulative opioid consumption by 24% (43 ± 24.1 mg versus 57 ± 28 mg, mean ± SD, p=0.02), translating into improved benefit of analgesia (OBAS: 2(0/4) versus 3(1/5.25), p=0.01), pain severity (m-BPI-sf: 1(1/2) versus 2(2/3), p < 0.01) and pain interference (m-BPI-sf: 1(0/1) versus 1(1/3), p=0.001), as well as reduced opioid-related side effects (OR-SDS score: 0.3(0.075/0.51) versus 0.4(0.2/0.83), p=0.03). Blood loss was significantly higher at 24 hours following surgery in the parecoxib group (4.3 gâ dL(-1) (3.6/4.9) versus (3.2 gâ dL(-1) (2.4/4.95), p=0.02). CONCLUSIONS: Following major abdominal surgery, parecoxib significantly improves patients' perceived analgesia. Parecoxib may however increase perioperative blood loss. Further trials are needed to evaluate the effects of selective cyclooxygenase-2 inhibitors on blood loss. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT00346268.
Subject(s)
Isoxazoles/therapeutic use , Pain, Postoperative/drug therapy , Postoperative Hemorrhage/chemically induced , Prostatectomy/adverse effects , Analgesia, Patient-Controlled/psychology , Analgesics, Opioid/adverse effects , Analgesics, Opioid/therapeutic use , Cyclooxygenase 2 Inhibitors/adverse effects , Cyclooxygenase 2 Inhibitors/therapeutic use , Double-Blind Method , Drug Therapy, Combination/adverse effects , Humans , Isoxazoles/adverse effects , Male , Middle Aged , Morphine/adverse effects , Morphine/therapeutic use , Pain Management , Pain Measurement , Pain, Postoperative/psychology , Patient SatisfactionABSTRACT
Neuromuscular blocking agents are used to facilitate tracheal intubation in patients undergoing ambulatory surgery. The use of high-dose neuromuscular blocking agents to achieve muscle paralysis throughout the case carries an increased risk of residual post-operative neuromuscular blockade, which is associated with increased respiratory morbidity. Visually monitoring the train-of-four (TOF) fade is not sensitive enough to detect a TOF fade between 0.4 and 0.9. A ratio <0.9 indicates inadequate recovery. Quantitative neuromuscular transmission monitoring (e.g., acceleromyography) should be used to exclude residual neuromuscular blockade at the end of the case. Residual neuromuscular blockade needs to be reversed with neostigmine, but it's use must be guided by TOF monitoring results since deep block cannot be reversed, and neostigmine administration after complete recovery of the TOF-ratio can induce muscle weakness. The development and use of new selectively binding reversal agents (sugammadex and calabadion) warrants reevaluation of this area of clinical practice.