Negation recognition in clinical natural language processing using a combination of the NegEx algorithm and a convolutional neural network.

BACKGROUND: Important clinical information of patients is present in unstructured free-text fields of Electronic Health Records (EHRs). While this information can be extracted using clinical Natural Language Processing (cNLP), the recognition of negation modifiers represents an important challenge. A wide range of cNLP applications have been developed to detect the negation of medical entities in clinical free-text, however, effective solutions for languages other than English are scarce. This study aimed at developing a solution for negation recognition in Spanish EHRs based on a combination of a customized rule-based NegEx layer and a convolutional neural network (CNN). METHODS: Based on our previous experience in real world evidence (RWE) studies using information embedded in EHRs, negation recognition was simplified into a binary problem ('affirmative' vs. 'non-affirmative' class). For the NegEx layer, negation rules were obtained from a publicly available Spanish corpus and enriched with custom ones, whereby the CNN binary classifier was trained on EHRs annotated for clinical named entities (cNEs) and negation markers by medical doctors. RESULTS: The proposed negation recognition pipeline obtained precision, recall, and F1-score of 0.93, 0.94, and 0.94 for the 'affirmative' class, and 0.86, 0.84, and 0.85 for the 'non-affirmative' class, respectively. To validate the generalization capabilities of our methodology, we applied the negation recognition pipeline on EHRs (6,710 cNEs) from a different data source distribution than the training corpus and obtained consistent performance metrics for the 'affirmative' and 'non-affirmative' class (0.95, 0.97, and 0.96; and 0.90, 0.83, and 0.86 for precision, recall, and F1-score, respectively). Lastly, we evaluated the pipeline against two publicly available Spanish negation corpora, the IULA and NUBes, obtaining state-of-the-art metrics (1.00, 0.99, and 0.99; and 1.00, 0.93, and 0.96 for precision, recall, and F1-score, respectively). CONCLUSION: Negation recognition is a source of low precision in the retrieval of cNEs from EHRs' free-text. Combining a customized rule-based NegEx layer with a CNN binary classifier outperformed many other current approaches. RWE studies highly benefit from the correct recognition of negation as it reduces false positive detections of cNE which otherwise would undoubtedly reduce the credibility of cNLP systems.

The Effect of Clusters of Double Triggering and Ineffective Efforts in Critically Ill Patients.

OBJECTIVES: To characterize clusters of double triggering and ineffective inspiratory efforts throughout mechanical ventilation and investigate their associations with mortality and duration of ICU stay and mechanical ventilation. DESIGN: Registry-based, real-world study. BACKGROUND: Asynchronies during invasive mechanical ventilation can occur as isolated events or in clusters and might be related to clinical outcomes. SUBJECTS: Adults requiring mechanical ventilation greater than 24 hours for whom greater than or equal to 70% of ventilator waveforms were available. INTERVENTIONS: We identified clusters of double triggering and ineffective inspiratory efforts and determined their power and duration. We used Fine-Gray's competing risk model to analyze their effects on mortality and generalized linear models to analyze their effects on duration of mechanical ventilation and ICU stay. MEASUREMENTS AND MAIN RESULTS: We analyzed 58,625,796 breaths from 180 patients. All patients had clusters (mean/d, 8.2 [5.4-10.6]; mean power, 54.5 [29.6-111.4]; mean duration, 20.3 min [12.2-34.9 min]). Clusters were less frequent during the first 48 hours (5.5 [2.5-10] vs 7.6 [4.4-9.9] in the remaining period [p = 0.027]). Total number of clusters/d was positively associated with the probability of being discharged alive considering the total period of mechanical ventilation (p = 0.001). Power and duration were similar in the two periods. Power was associated with the probability of being discharged dead (p = 0.03), longer mechanical ventilation (p < 0.001), and longer ICU stay (p = 0.035); cluster duration was associated with longer ICU stay (p = 0.027). CONCLUSIONS: Clusters of double triggering and ineffective inspiratory efforts are common. Although higher numbers of clusters might indicate better chances of survival, clusters with greater power and duration indicate a risk of worse clinical outcomes.

Efficacy and Safety of Taulinoplasty Compared with the Minimally Invasive Repair of Pectus Excavatum Approach to Correct Pectus Excavatum.

Background: Minimally invasive repair of pectus excavatum (MIRPE) technique is the current most used surgical method for pectus excavatum (PE) correction. To avoid MIRPE-required invasion of the mediastinum or pleural cavity, we developed taulinoplasty approach as an alternative option for the reduction of the sunken chest. Materials and Methods: This is a single-center unblind nonrandomized prospective pilot study, involving 26 MIRPE and 23 taulinoplasty patients. The primary safety endpoint was the presence of surgical complications, and the primary efficacy endpoint was Haller index measured postoperatively in taulinoplasty patients. Secondary endpoints were intensive care unit (ICU) and hospital length of stay (LOS), duration of surgical procedure, and postoperative pain management. Chi-square, Mann-Whitney, or Student "t" tests were used for comparison as appropriate. Results: There were no difference in median age (14.5 and 14 years), Haller index (4.63 and 4.17), or comorbidities. MIRPE and taulinoplasty procedures showed no difference regarding major or minor surgical complications. The efficacy endpoint of taulinoplasty was measured 5.5 (1.5-12) months after the procedure, exhibiting a significant reduction in the Haller index [4.17 (3.7-4.7) at baseline and 3.7 (2.9-4.1) postoperatively, P = .03]. Taulinoplasty required less surgical time (60.4 ± 15.5 versus 70.7 ± 15.9 minutes, P < .01); shorter ICU and hospital LOS (P < .01, respectively); and required fewer days of peridural, intravenous, and oral analgesia (P < .01, respectively). Conclusions: In this pilot study, taulinoplasty technique was as safe as the MIRPE approach and effective for the correction of PE, reducing surgical time, ICU and hospital LOS, and achieving better postoperative pain control.

Longitudinal Changes in Patient-Ventilator Asynchronies and Respiratory System Mechanics Before and After Tracheostomy.

BACKGROUND: This was a pilot study to analyze the effects of tracheostomy on patient-ventilator asynchronies and respiratory system mechanics. Data were extracted from an ongoing prospective, real-world database that stores continuous output from ventilators and bedside monitors. Twenty adult subjects were on mechanical ventilation and were tracheostomized during an ICU stay: 55% were admitted to the ICU for respiratory failure and 35% for neurologic conditions; the median duration of mechanical ventilation before tracheostomy was 12 d; and the median duration of mechanical ventilation was 16 d. METHODS: We compared patient-ventilator asynchronies (the overall asynchrony index and the rates of specific asynchronies) and respiratory system mechanics (respiratory-system compliance and airway resistance) during the 24 h before tracheostomy versus the 24 h after tracheostomy. We analyzed possible differences in these variables among the subjects who underwent surgical versus percutaneous tracheostomy. To compare longitudinal changes in the variables, we used linear mixed-effects models for repeated measures along time in different observation periods. A total of 920 h of mechanical ventilation were analyzed. RESULTS: Respiratory mechanics and asynchronies did not differ significantly between the 24-h periods before and after tracheostomy: compliance of the respiratory system median (IQR) (47.9 [41.3 - 54.6] mL/cm H2O vs 47.6 [40.9 - 54.3] mL/cm H2O; P = .94), airway resistance (9.3 [7.5 - 11.1] cm H2O/L/s vs 7.0 [5.2 - 8.8] cm H2O/L/s; P = .07), asynchrony index (2.0% [1.1 - 3.6%] vs 4.1% [2.3 - 7.6%]; P = .09), ineffective expiratory efforts (0.9% [0.4 - 1.8%] vs 2.2% [1.0 - 4.4%]; P = .08), double cycling (0.5% [0.3 - 1.0%] vs 0.9% [0.5 - 1.9%]; P = .24), and percentage of air trapping (7.6% [4.2 - 13.8%] vs 10.6% [5.9 - 19.2%]; P = .43). No differences in respiratory mechanics or patient-ventilator asynchronies were observed between percutaneous and surgical procedures. CONCLUSIONS: Tracheostomy did not affect patient-ventilator asynchronies or respiratory mechanics within 24 h before and after the procedure.

Publisher Correction: Development and validation of a sample entropy-based method to identify complex patient-ventilator interactions during mechanical ventilation.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Cognitive phenotypes 1 month after ICU discharge in mechanically ventilated patients: a prospective observational cohort study.

BACKGROUND: ICU patients undergoing invasive mechanical ventilation experience cognitive decline associated with their critical illness and its management. The early detection of different cognitive phenotypes might reveal the involvement of diverse pathophysiological mechanisms and help to clarify the role of the precipitating and predisposing factors. Our main objective is to identify cognitive phenotypes in critically ill survivors 1 month after ICU discharge using an unsupervised machine learning method, and to contrast them with the classical approach of cognitive impairment assessment. For descriptive purposes, precipitating and predisposing factors for cognitive impairment were explored. METHODS: A total of 156 mechanically ventilated critically ill patients from two medical/surgical ICUs were prospectively studied. Patients with previous cognitive impairment, neurological or psychiatric diagnosis were excluded. Clinical variables were registered during ICU stay, and 100 patients were cognitively assessed 1 month after ICU discharge. The unsupervised machine learning K-means clustering algorithm was applied to detect cognitive phenotypes. Exploratory analyses were used to study precipitating and predisposing factors for cognitive impairment. RESULTS: K-means testing identified three clusters (K) of patients with different cognitive phenotypes: K1 (n = 13), severe cognitive impairment in speed of processing (92%) and executive function (85%); K2 (n = 33), moderate-to-severe deficits in learning-memory (55%), memory retrieval (67%), speed of processing (36.4%) and executive function (33.3%); and K3 (n = 46), normal cognitive profile in 89% of patients. Using the classical approach, moderate-to-severe cognitive decline was recorded in 47% of patients, while the K-means method accurately classified 85.9%. The descriptive analysis showed significant differences in days (p = 0.016) and doses (p = 0.039) with opioid treatment in K1 vs. K2 and K3. In K2, there were more women, patients were older and had more comorbidities (p = 0.001) than in K1 or K3. Cognitive reserve was significantly (p = 0.001) higher in K3 than in K1 or K2. CONCLUSION: One month after ICU discharge, three groups of patients with different cognitive phenotypes were identified through an unsupervised machine learning method. This novel approach improved the classical classification of cognitive impairment in ICU survivors. In the exploratory analysis, gender, age and the level of cognitive reserve emerged as relevant predisposing factors for cognitive impairment in ICU patients. TRIAL REGISTRATION: ClinicalTrials.gov Identifier:NCT02390024; March 17,2015.

Development and validation of a sample entropy-based method to identify complex patient-ventilator interactions during mechanical ventilation.

Patient-ventilator asynchronies can be detected by close monitoring of ventilator screens by clinicians or through automated algorithms. However, detecting complex patient-ventilator interactions (CP-VI), consisting of changes in the respiratory rate and/or clusters of asynchronies, is a challenge. Sample Entropy (SE) of airway flow (SE-Flow) and airway pressure (SE-Paw) waveforms obtained from 27 critically ill patients was used to develop and validate an automated algorithm for detecting CP-VI. The algorithm's performance was compared versus the gold standard (the ventilator's waveform recordings for CP-VI were scored visually by three experts; Fleiss' kappa = 0.90 (0.87-0.93)). A repeated holdout cross-validation procedure using the Matthews correlation coefficient (MCC) as a measure of effectiveness was used for optimization of different combinations of SE settings (embedding dimension, m, and tolerance value, r), derived SE features (mean and maximum values), and the thresholds of change (Th) from patient's own baseline SE value. The most accurate results were obtained using the maximum values of SE-Flow (m = 2, r = 0.2, Th = 25%) and SE-Paw (m = 4, r = 0.2, Th = 30%) which report MCCs of 0.85 (0.78-0.86) and 0.78 (0.78-0.85), and accuracies of 0.93 (0.89-0.93) and 0.89 (0.89-0.93), respectively. This approach promises an improvement in the accurate detection of CP-VI, and future study of their clinical implications.

Gas exchange and lung mechanics in patients with acute respiratory distress syndrome: comparison of three different strategies of positive end expiratory pressure selection.

PURPOSE: The purpose of the study was to compare gas exchange and lung mechanics between different strategies to select positive end-expiratory pressure (PEEP) in acute respiratory distress syndrome (ARDS). METHODS: In 20 consecutive ARDS patients, 3 PEEP selection strategies were evaluated. One strategy was based on oxygenation using the ARDS network PEEP/fraction of inspired oxygen (Fio2) table; and two were based on lung mechanics, either PEEP titrated to reach a plateau pressure of 28 to 30 cm H2O as in the ExPress trial or best respiratory compliance method during a derecruitment maneuver. Gas exchange, airway pressures, stress index (SI), and end-expiratory transpulmonary pressure (P(tpe)) and end-inspiratory transpulmonary pressure (P(tpi)) values were assessed. Data are expressed as median (interquartile range [IQR]). RESULTS: Lower total PEEP levels were observed with the use of the PEEP/Fio2 table (8.7 [6-10] cm H2O); intermediate PEEP levels, with the Best Compliance approach (13.0 [10.2-13.8] cm H2O); and higher PEEP levels, with the ExPress strategy (16.5 [15.0-18.5] cm H2O) (P < .01). Pao2/Fio2 ratio was lower with the PEEP/Fio2 table. Oxygenation with Best Compliance approach and ExPress strategy was not different with lower plateau pressure in the former (23 [20-25] vs 30 [29-30] cm H2O; P < .01). Paco2 was slightly higher with the ExPress method than the others 2 strategies. Negative P(tpe) was observed in 35% of the patients with the PEEP/Fio2 table, in 15% applying the Best Compliance, and in only 1 case with the ExPress method. Higher SI and P(tpi), with lower lung compliance, were obtained with ExPress strategy. CONCLUSIONS: Using a best respiratory compliance approach resulted in better oxygenation levels without risk of overdistension according to SI and P(tpi), achieving a mild risk of lung collapse according to P(tpe).

[Severe respiratory disease in an intensive care unit during influenza A(H1N1) 2009 pandemia]. / Enfermedad respiratoria grave en terapia intensiva durante la pandemia por el virus de influenza A (H1N1) 2009.

We describe characteristics of patients admitted to our intensive care unit with severe acute respiratory illness and influenza-like syndrome during the first months of the pandemic influenza A(H1N1) 2009 in Argentina. We analyzed clinical data, severity scores, laboratory tests, microbiological and radiological findings at admission, clinical outcomes and in-hospital mortality. H1N1 was confirmed by RT-PCR. Data from positive and negative PCR patients were compared. We admitted 31 adult patients between June and July 2009; median age: 54 years (IQR 33-66). A 54% (17) had positive PCR; 16 patients presented underlying medical conditions. Bilateral interstitial opacities were observed in chest radiography in 20 cases; 5 had unilateral lobar consolidation. Bacterial co-infection (isolation or IgM antibodies for bacterial infections) was found in 21 patients. Mechanical ventilation was required in 23 patients and 18 developed ARDS. Lymphopenia and increased creatine kinase levels were frequently observed (83% and 65% among PCR+ and PCR- respectively). Six patients died (19%); they were all over 75 years old, had cancer or immune-suppression. Early antiviral treatment (≤ 48 hours from symptoms onset) was associated with less frequency of mechanical ventilation (54% vs. 89%, p: 0.043). There were no differences in analyzed variables when comparing H1N1 positive and H1N1 negative patients; which suggests this approach as a most correct in future epidemic outbreaks. H1N1 infection was associated with severe respiratory illness and ARDS. Fatal outcome was observed in very old patients, or in those with major co-morbidities.

Enfermedad respiratoria grave en terapia intensiva durante la pandemia por el virus de influenza A (H1N1) 2009 / Severe respiratory disease in an intensive care unit during influenza A(H1N1)2009 pandemia

Se describen pacientes hospitalizados en una unidad de terapia intensiva por enfermedad respiratoria aguda grave con características de influenza durante los primeros meses de la pandemia por influenza A(H1N1) 2009 en la Argentina. Evaluamos datos clínicos, scores de gravedad, pruebas de laboratorio, microbiología y radiología torácica al ingreso, evolución y mortalidad hospitalaria, comparando pacientes con y sin confirmación de H1N1 por test de reacción de polimerasa en cadena, transcriptasa reversa (RT-PCR). Entre junio y julio de 2009 se internaron 31 pacientes adultos con una mediana de edad de 54 años (percentilo 25-75: 33-66). Presentaron test positivo para H1N1, 17 pacientes. Tenían al menos una condición concurrente 16 pacientes. La expresión radiográfica más frecuente fue infiltrados intersticio-alveolares bilaterales en 20 casos; 5 tenían consolidación lobar unilateral. La coinfección bacteriana (aislamiento de bacterias o IgM positiva para infecciones bacterianas), se demostró en 21 pacientes. Requirieron ventilación mecánica 23 pacientes y 18 desarrollaron síndrome de distrés respiratorio agudo (SDRA). La linfopenia y elevación de creatinina-fosfoquinasa fue frecuente (83% y 65%, respectivamente). Los 6 pacientes que murieron (19%) eran mayores de 75 años o tenían cáncer o inmunodepresión. El tratamiento antiviral temprano (≤ 48 horas) se asoció a menor necesidad de ventilación mecánica (54% vs. 89%; p: 0.043). No hubo diferencia significativa en las variables analizadas entre el grupo H1N1 positivo y el negativo, lo que sugiere tener igual enfoque terapéutico frente a una epidemia. La infección por H1N1 determinó falla respiratoria aguda y SDRA. La mortalidad ocurrió en pacientes añosos o con co-morbilidades graves.

We describe characteristics of patients admitted to our intensive care unit with severe acute respiratory illness and influenza-like syndrome during the first months of the pandemic influenza A(H1N1) 2009 in Argentina. We analyzed clinical data, severity scores, laboratory tests, microbiological and radiological findings at admission, clinical outcomes and in-hospital mortality. H1N1 was confirmed by RT-PCR. Data from positive and negative PCR patients were compared. We admitted 31 adult patients between June and July 2009; median age: 54 years (IQR 33-66). A 54% (17) had positive PCR; 16 patients presented underlying medical conditions. Bilateral interstitial opacities were observed in chest radiography in 20 cases; 5 had unilateral lobar consolidation. Bacterial co-infection (isolation or IgM antibodies for bacterial infections) was found in 21 patients. Mechanical ventilation was required in 23 patients and 18 developed ARDS. Lymphopenia and increased creatine kinase levels were frequently observed (83% and 65% among PCR+ and PCR- respectively). Six patients died (19%); they were all over 75 years old, had cancer or immune-suppression. Early antiviral treatment (≤ 48 hours from symptoms onset) was associated with less frequency of mechanical ventilation (54% vs. 89%, p: 0.043). There were no differences in analyzed variables when comparing H1N1 positive and H1N1 negative patients; which suggests this approach as a most correct in future epidemic outbreaks. H1N1 infection was associated with severe respiratory illness and ARDS. Fatal outcome was observed in very old patients, or in those with major co-morbidities.

Erythroid expansion and survival in response to acute anemia stress: the role of EPO receptor, GATA-1, Bcl-xL and caspase-3.

Erythropoietic stress occurs under conditions of tissular hypoxia, such as anemia. Functional relationships between erythroid bone marrow (BM) proliferation, differentiation, the expression of survival and apoptotic related proteins, as well as the features of the BM microenvironment upon acute anemic stress, are not fully elucidated. To achieve this aim, CF-1 Swiss mice were injected with a single dose of 5-fluorouracil (5-FU, 150 mg/kg ip) and a multiparametric analysis was conducted for 20 days. Apoptosis (TUNEL assay), BM architecture organization (scanning electronic microscopy), proliferation (DNA assay), differentiation (clonogenic cultures), expression of survival erythroid related proteins (EPO-R, GATA-1, Bcl-xL) as well as the expression of apoptotic- related proteins (Bax, activated Caspase-3) by Western blotting, were evaluated. Experimental data showed that apoptosis, arrest of cell proliferation and disruptions of BM architecture were maximal within the first period of acute stress (1-3 days). Bax and caspase-3 overexpressions were also coincident during this acute period. Moreover, from day 5 upon drug challenge BM responds to acute stress through the EPO-EPO-R system, prompting expressions of GATA-1 and Bcl-xL. Erythroid proliferation rates and red-cell-committed progenitors enhanced in a coordinated way to restore the size and function of the red cell compartment. A second overexpression wave of active caspase-3 was noticed during stress recovery. Together, these results indicate that in response to acute stress a dramatic increase in CFU-E (erythroid colony forming units) population is concomitant with upregulation of EPO-R, GATA-1 and Bcl-xL in the BM erythroid compartment, and that these concurrent processes are crucial for acquiring proper erythroid cell functionality without delayed response to tissular hypoxia.

In vivo 5-fluorouracil-[corrected]induced apoptosis on murine thymocytes: involvement of FAS, Bax and Caspase3.

Apoptosis is a highly regulated and programmed cell breakdown process characterized by numerous changes. It was reported as the major mechanism of anticancer drug-induced cells death. Unfortunately, many of these drugs are non-specific and cause severe side effects. The effects of 5-fluorouracil (5-FU) on the apoptotic events in normal murine thymus were evaluated using an in vivo model. A single dose of 5-FU (150 mg/kg ip) was injected to CF-1 mice. A multiparametric analysis of thymic weight, cellularity, viability, architectural organization, apoptosis, DNA fragmentation, and the expression of several apoptotic proteins was evaluated in 10 days time-course study post-5-FU dosing. Total organ weights, thymocyte counts, and cell viabilities diminished drastically from the second day. The thymus architecture assessed through electron scanning microscopy revealed deep alterations and the lost of cell-cell contact between the first and the third days. DNA fragmentation and apoptotic indexes (May Grünwald Giemsa staining, double fluorescent dyes, and TdT-mediated dUTP nick-end labeling assay) revealed that cell death was maximal on the second day (three times over control). Furthermore, the pro-apoptotic proteins FAS and Bax were strongly up-regulated during the first 2 days. The aforementioned morphological and biochemical changes were also accompanied within the same period by caspase 3 activation. This study revealed that in vivo apoptosis in normal thymus after 5-FU administration is related to FAS, Bax, and Caspase 3 co-expressions under the current experimental conditions, these findings, therefore, contribute to a new insight into the molecular mechanisms involved during 5-FU administration upon the thymus and the possible events committed in the lymphophenia associated with chemotherapy.