Generalized Confidence Intervals for Ratios of Standard Deviations Based on Log-Normal Distribution when Times Follow Weibull Distributions.

Modern anesthetic drugs ensure the efficacy of general anesthesia. Goals include reducing variability in surgical, tracheal extubation, post-anesthesia care unit, or intraoperative response recovery times. Generalized confidence intervals based on the log-normal distribution compare variability between groups, specifically ratios of standard deviations. The alternative statistical approaches, performing robust variance comparison tests, give P-values, not point estimates nor confidence intervals for the ratios of the standard deviations. We performed Monte-Carlo simulations to learn what happens to confidence intervals for ratios of standard deviations of anesthesia-associated times when analyses are based on the log-normal, but the true distributions are Weibull. We used simulation conditions comparable to meta-analyses of most randomized trials in anesthesia, n ≈ 25 and coefficients of variation ≈ 0.30 . The estimates of the ratios of standard deviations were positively biased, but slightly, the ratios being 0.11% to 0.33% greater than nominal. In contrast, the 95% confidence intervals were very wide (i.e., > 95% of P ≥ 0.05). Although substantive inferentially, the differences in the confidence limits were small from a clinical or managerial perspective, with a maximum absolute difference in ratios of 0.016. Thus, P < 0.05 is reliable, but investigators should plan for Type II errors at greater than nominal rates.

Training a Deep Contextualized Language Model for International Classification of Diseases, 10th Revision Classification via Federated Learning: Model Development and Validation Study.

BACKGROUND: The automatic coding of clinical text documents by using the International Classification of Diseases, 10th Revision (ICD-10) can be performed for statistical analyses and reimbursements. With the development of natural language processing models, new transformer architectures with attention mechanisms have outperformed previous models. Although multicenter training may increase a model's performance and external validity, the privacy of clinical documents should be protected. We used federated learning to train a model with multicenter data, without sharing data per se. OBJECTIVE: This study aims to train a classification model via federated learning for ICD-10 multilabel classification. METHODS: Text data from discharge notes in electronic medical records were collected from the following three medical centers: Far Eastern Memorial Hospital, National Taiwan University Hospital, and Taipei Veterans General Hospital. After comparing the performance of different variants of bidirectional encoder representations from transformers (BERT), PubMedBERT was chosen for the word embeddings. With regard to preprocessing, the nonalphanumeric characters were retained because the model's performance decreased after the removal of these characters. To explain the outputs of our model, we added a label attention mechanism to the model architecture. The model was trained with data from each of the three hospitals separately and via federated learning. The models trained via federated learning and the models trained with local data were compared on a testing set that was composed of data from the three hospitals. The micro F1 score was used to evaluate model performance across all 3 centers. RESULTS: The F1 scores of PubMedBERT, RoBERTa (Robustly Optimized BERT Pretraining Approach), ClinicalBERT, and BioBERT (BERT for Biomedical Text Mining) were 0.735, 0.692, 0.711, and 0.721, respectively. The F1 score of the model that retained nonalphanumeric characters was 0.8120, whereas the F1 score after removing these characters was 0.7875-a decrease of 0.0245 (3.11%). The F1 scores on the testing set were 0.6142, 0.4472, 0.5353, and 0.2522 for the federated learning, Far Eastern Memorial Hospital, National Taiwan University Hospital, and Taipei Veterans General Hospital models, respectively. The explainable predictions were displayed with highlighted input words via the label attention architecture. CONCLUSIONS: Federated learning was used to train the ICD-10 classification model on multicenter clinical text while protecting data privacy. The model's performance was better than that of models that were trained locally.

Risk factors for delirium in elderly patients after lumbar spinal fusion.

OBJECTIVE: To identify perioperative risk factors for postoperative delirium (POD) in patients aged 65 or older undergoing lumbar spinal fusion procedures. PATIENTS AND METHODS: A retrospective cohort analysis was performed on patients undergoing lumbar spinal fusion over an approximately three-year period at a single institution. Demographic and perioperative data were obtained from electronic medical records. The primary outcome was the presence of postoperative delirium assayed by the Delirium Observation Screening Scale (DOSS) and Confusion Assessment Method for the ICU (CAM-ICU). Univariate and multivariate analyses were performed on the data. RESULTS: Of the 702 patients included in the study, 173 (24.6%) developed POD. Our analysis revealed that older age (p < 0.001), lower preoperative hemoglobin (p < 0.001), and higher ASA status (p < 0.001), were significant preoperative risk factors for developing POD. The only significant intraoperative risk factor was a higher number of spinal levels that were instrumented (p < 0.001). Higher pain scores on postoperative day 1 (p < 0.001), and lower postoperative hemoglobin (p < 0.001) were associated with increased POD; as were ICU admission (p < 0.001) and increased length of ICU stay (p < 0.001). Patients who developed POD had a longer hospital stay (p < 0.001) with lower rates of discharge to home as opposed to an inpatient facility (p < 0.001). CONCLUSION: Risk factors for POD in older adults undergoing lumbar spinal fusion surgery include advanced age, diabetes, lower preoperative and postoperative hemoglobin, higher ASA grade, greater extent of surgery, and higher postoperative pain scores. Patients with delirium had a higher incidence of postoperative ICU admission, increased length of stay, decreased likelihood of discharge to home and increased mortality, all consistent with prior studies. Further studies may determine whether adequate management of anemia and pain lead to a reduction in the incidence of postoperative delirium in these patients.

Automatic International Classification of Diseases Coding System: Deep Contextualized Language Model With Rule-Based Approaches.

BACKGROUND: The tenth revision of the International Classification of Diseases (ICD-10) is widely used for epidemiological research and health management. The clinical modification (CM) and procedure coding system (PCS) of ICD-10 were developed to describe more clinical details with increasing diagnosis and procedure codes and applied in disease-related groups for reimbursement. The expansion of codes made the coding time-consuming and less accurate. The state-of-the-art model using deep contextual word embeddings was used for automatic multilabel text classification of ICD-10. In addition to input discharge diagnoses (DD), the performance can be improved by appropriate preprocessing methods for the text from other document types, such as medical history, comorbidity and complication, surgical method, and special examination. OBJECTIVE: This study aims to establish a contextual language model with rule-based preprocessing methods to develop the model for ICD-10 multilabel classification. METHODS: We retrieved electronic health records from a medical center. We first compared different word embedding methods. Second, we compared the preprocessing methods using the best-performing embeddings. We compared biomedical bidirectional encoder representations from transformers (BioBERT), clinical generalized autoregressive pretraining for language understanding (Clinical XLNet), label tree-based attention-aware deep model for high-performance extreme multilabel text classification (AttentionXLM), and word-to-vector (Word2Vec) to predict ICD-10-CM. To compare different preprocessing methods for ICD-10-CM, we included DD, medical history, and comorbidity and complication as inputs. We compared the performance of ICD-10-CM prediction using different preprocesses, including definition training, external cause code removal, number conversion, and combination code filtering. For the ICD-10 PCS, the model was trained using different combinations of DD, surgical method, and key words of special examination. The micro F1 score and the micro area under the receiver operating characteristic curve were used to compare the model's performance with that of different preprocessing methods. RESULTS: BioBERT had an F1 score of 0.701 and outperformed other models such as Clinical XLNet, AttentionXLM, and Word2Vec. For the ICD-10-CM, the model had an F1 score that significantly increased from 0.749 (95% CI 0.744-0.753) to 0.769 (95% CI 0.764-0.773) with the ICD-10 definition training, external cause code removal, number conversion, and combination code filter. For the ICD-10-PCS, the model had an F1 score that significantly increased from 0.670 (95% CI 0.663-0.678) to 0.726 (95% CI 0.719-0.732) with a combination of discharge diagnoses, surgical methods, and key words of special examination. With our preprocessing methods, the model had the highest area under the receiver operating characteristic curve of 0.853 (95% CI 0.849-0.855) and 0.831 (95% CI 0.827-0.834) for ICD-10-CM and ICD-10-PCS, respectively. CONCLUSIONS: The performance of our model with the pretrained contextualized language model and rule-based preprocessing method is better than that of the state-of-the-art model for ICD-10-CM or ICD-10-PCS. This study highlights the importance of rule-based preprocessing methods based on coder coding rules.

Healing process of Type II odontoid fractures after C1-C2 posterior screw fixation: Predictive factors for pseudoarthrosis.

Background: The healing process after C1-C2 posterior screw fixation (C1-C2 PSF) for odontoid fractures is not well understood. Here, we evaluated such processes and identified factors potentially contributing to pseudoarthroses following fusions for Type II odontoid fractures. Methods: Pre- and post-operative cervical radiographs and computed tomography (CT) images from 15 patients with preoperative Type II odontoid fractures who underwent C1-C2 PSF were retrospectively reviewed. Results: CT images identified three areas of bone fusion: The primary fracture site in the dens (9/15 patients, 60%), the atlanto-dental interspace (ADI) (10/15, 67%), and C1-C2 interlaminar space after onlay bone grafting (4/15, 27%). All patients showed bone fusion in at least one of three areas, while only one patient (6.7%) achieved bone fusion in all three areas. With these overall criteria, nine of 15 patients (60%) were considered fused, while six patients (40%) were determined to exhibit pseudoarthroses. Univariate analyzes showed that the preoperative C2-C7 SVA for the nonunion group was significantly larger versus the union group, and bone fusion at the level of the ADI was significantly more common in the nonunion versus the union group. Conclusion: CT studies identified three anatomical areas where bone fusion likely occurs after C1-C2 PSF. Increased sagittal balance in the cervical spine may negatively impact the fusion of odontoid fractures. Further, bone fusion occurring at other sites, not the primary fracture location, through stress shielding may contribute to delayed or failed fusions.

Predicting Postoperative Mortality With Deep Neural Networks and Natural Language Processing: Model Development and Validation.

BACKGROUND: Machine learning (ML) achieves better predictions of postoperative mortality than previous prediction tools. Free-text descriptions of the preoperative diagnosis and the planned procedure are available preoperatively. Because reading these descriptions helps anesthesiologists evaluate the risk of the surgery, we hypothesized that deep learning (DL) models with unstructured text could improve postoperative mortality prediction. However, it is challenging to extract meaningful concept embeddings from this unstructured clinical text. OBJECTIVE: This study aims to develop a fusion DL model containing structured and unstructured features to predict the in-hospital 30-day postoperative mortality before surgery. ML models for predicting postoperative mortality using preoperative data with or without free clinical text were assessed. METHODS: We retrospectively collected preoperative anesthesia assessments, surgical information, and discharge summaries of patients undergoing general and neuraxial anesthesia from electronic health records (EHRs) from 2016 to 2020. We first compared the deep neural network (DNN) with other models using the same input features to demonstrate effectiveness. Then, we combined the DNN model with bidirectional encoder representations from transformers (BERT) to extract information from clinical texts. The effects of adding text information on the model performance were compared using the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC). Statistical significance was evaluated using P<.05. RESULTS: The final cohort contained 121,313 patients who underwent surgeries. A total of 1562 (1.29%) patients died within 30 days of surgery. Our BERT-DNN model achieved the highest AUROC (0.964, 95% CI 0.961-0.967) and AUPRC (0.336, 95% CI 0.276-0.402). The AUROC of the BERT-DNN was significantly higher compared to logistic regression (AUROC=0.952, 95% CI 0.949-0.955) and the American Society of Anesthesiologist Physical Status (ASAPS AUROC=0.892, 95% CI 0.887-0.896) but not significantly higher compared to the DNN (AUROC=0.959, 95% CI 0.956-0.962) and the random forest (AUROC=0.961, 95% CI 0.958-0.964). The AUPRC of the BERT-DNN was significantly higher compared to the DNN (AUPRC=0.319, 95% CI 0.260-0.384), the random forest (AUPRC=0.296, 95% CI 0.239-0.360), logistic regression (AUPRC=0.276, 95% CI 0.220-0.339), and the ASAPS (AUPRC=0.149, 95% CI 0.107-0.203). CONCLUSIONS: Our BERT-DNN model has an AUPRC significantly higher compared to previously proposed models using no text and an AUROC significantly higher compared to logistic regression and the ASAPS. This technique helps identify patients with higher risk from the surgical description text in EHRs.

Automatic ICD-10 Coding and Training System: Deep Neural Network Based on Supervised Learning.

BACKGROUND: The International Classification of Diseases (ICD) code is widely used as the reference in medical system and billing purposes. However, classifying diseases into ICD codes still mainly relies on humans reading a large amount of written material as the basis for coding. Coding is both laborious and time-consuming. Since the conversion of ICD-9 to ICD-10, the coding task became much more complicated, and deep learning- and natural language processing-related approaches have been studied to assist disease coders. OBJECTIVE: This paper aims at constructing a deep learning model for ICD-10 coding, where the model is meant to automatically determine the corresponding diagnosis and procedure codes based solely on free-text medical notes to improve accuracy and reduce human effort. METHODS: We used diagnosis records of the National Taiwan University Hospital as resources and apply natural language processing techniques, including global vectors, word to vectors, embeddings from language models, bidirectional encoder representations from transformers, and single head attention recurrent neural network, on the deep neural network architecture to implement ICD-10 auto-coding. Besides, we introduced the attention mechanism into the classification model to extract the keywords from diagnoses and visualize the coding reference for training freshmen in ICD-10. Sixty discharge notes were randomly selected to examine the change in the F1-score and the coding time by coders before and after using our model. RESULTS: In experiments on the medical data set of National Taiwan University Hospital, our prediction results revealed F1-scores of 0.715 and 0.618 for the ICD-10 Clinical Modification code and Procedure Coding System code, respectively, with a bidirectional encoder representations from transformers embedding approach in the Gated Recurrent Unit classification model. The well-trained models were applied on the ICD-10 web service for coding and training to ICD-10 users. With this service, coders can code with the F1-score significantly increased from a median of 0.832 to 0.922 (P<.05), but not in a reduced interval. CONCLUSIONS: The proposed model significantly improved the F1-score but did not decrease the time consumed in coding by disease coders.

Possible Association of IFN-γ Gene -316A/G SNP with Humoral Immune Response to Killed H5N1 HPAI Vaccine in a Red Junglefowl Population.

To understand the role of interferon (IFN)-γ gene (IFNG) single-nucleotide polymorphisms (SNPs) in the resistance to H5N1 highly pathogenic avian influenza (HPAI), we determined the SNPs, the log2-transformed specific antibody titers, and ex vivo recall antigen-specific IFN-γ production by peripheral blood mononuclear cells (PBMCs) in 62 red junglefowls (Gallus gallus) immunized twice with inactivated H5N1 HPAI vaccine. Consequently, 52 SNPs were detected in the amplified 1137-bp length covering the promoter region and all exons, with +2133A/T SNP in the coding DNA sequence that caused a missense mutation and was identified in G. gallus for the first time. For -316A/G SNP genotypes, the chi-squared test showed that the bird sample was in the Hardy-Weinberg equilibrium (P=0.369>0.05), and the least squares analysis revealed an increasing tendency in the antibody titers with G to A substitution so that a significant difference occurred between the AA genotype (8.031±0.247) and the GG genotype (6.571±0.528) (P=0.015<0.05, 95% confidence interval [CI]: 0.0004-0.0866), as basically consistent with the antigen-induced IFN-γ protein expression, which indicated possible association of the -316A/G SNP with a secondary humoral immune response to the HPAI vaccine in the bird population. These findings may help to improve genetic resistance using cross-breeding and enhance HPAI vaccine-induced immunity in chicken production.

Rapid cloning and comparative sequence analysis of full-length cDNA of Rhesus monkey (Macaca mulatta) resistin.

Resistin protein is thought to link insulin resistance in murine models of obesity and type-2 diabetes, but the role of resistin in human studies of inflammatory metabolic disorders have generated conflicting data. Here, we describe the structure of the resistin gene using adipose tissue from non-human primates (NHPs), which have been used extensively to model a host of human diseases. Full-length cDNA from rhesus macaque resistin obtained by rapid amplification of cDNA ends (RACE) is comprised of 526 nucleotides covering an open-reading frame (ORF) that encodes a 108-amino-acid protein that is 92% homologous with the human counterpart but only 60% homologous with the murine counterpart. Using a modified polymerase chain reaction technique, we identified single nucleotide polymorphisms and a 78-bp deletion within resistin cDNA of nine rhesus macaques. Comparisons of the full-length cDNA sequence and an amplified 569-bp genomic DNA sequence revealed an error in published predictions arising from genomic studies about the gene's exon 3 region. Our data show, for the first time, the full-length macaque resistin cDNA sequence (GenBank: JF740676.1). These findings will illuminate future studies into the role of resistin in NHP models of inflammatory metabolic diseases.

Canine distemper virus utilizes different receptors to infect chicken embryo fibroblasts and vero cells.

Inducing animal viruses to adapt to chicken embryos or chicken embryo fibroblasts (CEF) is a common method to develop attenuated live vaccines with full security. Canine distemper virus (CDV) also does this, but the mechanisms and particular receptors remain unclear. Virus overlay protein blot assays were carried out on CEF membrane proteins, which were extracted respectively with a Mem-PER™ kit, a radioimmunoprecipitation assay buffer or a modified co-immunoprecipitation method, and revealed a common 57 kDa positive band that differed from the 42-kDa positive band in Vero cells and also from those receptors reported in lymphocytes and 293 cells, indicating a receptor diversity of CDV and the possibility of the 57-kDa protein acting as a receptor that is involved in adaptive infection of CDV Kunming strain to CEF.

Separation and purification of Escherichia coli-expressed human thymosin-α1 using affinity chromatography and high-performance liquid chromatography.

In this study, a human thymosin-α1 (hTα1) fusion protein was overexpressed in Escherichia coli (E. coli). The hexahistidine-tagged hTα1 fusion protein was obtained in soluble form in cells of the engineered E. coli strain BL21 (DE3)/pET-28a-hTα1 that had been induced with isopropyl -D-1-thiogalactopyranoside (IPTG). The recombinant protein accounted for approximately 50-60% of the total protein. We then developed and validated a separation method for hTα1 from E. coli cells based on thermal denaturation, nickel-resin affinity chromatography and high-performance liquid chromatography. The purification method showed good reproducibility and was easy to operate. Purified recombinant hTα1 of high homogeneity was characterized and found to be of high purity (over 99%), as determined by high-voltage electrophoresis and high-performance liquid chromatography analysis. Isoelectric focusing analysis indicated a pI of approximately 4.0, and full wavelength screening showed an optimal absorbance wavelength at around 214nm.

Liposomal plasmid DNA encoding human thymosin alpha and interferon omega potently inhibits liver tumor growth in ICR mice.

AIM: To evaluate the potential therapeutic effect of liposomal gene delivery, genes encoding for human thymosin alpha1 (Talpha1) and interferon omega1 were injected via the tail vein into mice bearing a Hep-A-22 liver tumor. METHODS: The cDNA of human Talpha1 and interferon omega1 were obtained by synthesis or reverse transcription-polymerase chain reaction (RT-PCR), respectively. Eukaryotic expressing vectors pIRES2, encoding Talpha1 and/or interferon omega1, were constructed and injected with liposome via the tail vein into ICR mice bearing a Hep-A-22 tumor. The potency of tumor inhibition was evaluated when three treated groups were compared with the group receiving the empty vector. Apoptosis of tumor cells was investigated by analyzing DNA fragmentation. RESULTS: Only the group treated with dual-gene plasmid reached an eligible level of tumor inhibition (43%). The difference in tumor weight was statistically significant between the Talpha1 gene or the interferon omega1 gene treated groups and the control (P<0.05), and highly significant between the dual-gene treated group and the control (P<0.01). DNA ladder was observed in the tumor cells from the purpose gene treated groups but not from the control. CONCLUSION: The dual-gene plasmid-liposome complex showed more potent inhibition than the single gene constructs on the growth of Hep-A-22 tumor cells in mice, which may be attributed to indirect and additive induction of apoptosis in tumor cells by increased expression of Talpha1 and interferon omega1.

Overexpression of soluble human thymosin alpha 1 in Escherichia coli.

Synthesized gene of human thymosin alpha 1 (Talpha1) was inserted into pET-28a, pET-9c, pThioHis B, pGEX-2T or pBV222 and then inductively expressed in strains of Escherichia coli. Among the five expression systems, the BL21/pET-28a system provides the highest expression level of fusion protein in a soluble form, which is up to 70% of total expressed bacterial proteins as visualized by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The resulting fusion protein purified through nickel affinity chromatography accounts for 2.53% of the wet bacterial pellet weight and reaches 94.5% purity by SDS-PAGE. These results indicate the potential of this expression system for high-throughput production of recombinant Talpha1.