Safety and efficacy of the FAKHRAVAC compared with BBIBP-Corv2 against SARS-CoV-2 in adults: a non-inferiority multi-center trial.

BACKGROUND: We compared Fakhravac and BBIBP-Corv2 vaccines in a phase III trial. METHOD: We conducted a multicenter, parallel-group, active-control, non-inferiority clinical trial with pragmatic considerations assessing the safety and efficacy of Fakhravac and BBIBP-Corv2 vaccines. We started with two randomized double-blind arms and added two non-randomized open-label arms (based on participant preference) because of slow recruitment. The adult population received 0.5 ml (10 µg per dose) intramuscular injections of Fakhravac or BBIBP-Corv-2 vaccines 21 days apart. The primary outcome was the occurrence of PCR-positive symptomatic Covid-19 disease 14 days or more after the second injection. A 10% non-inferiority margin to the reported 72.8% efficacy of BBIBP-Corv2 was assumed. Cox proportional hazard modeling was used to estimate hazard ratios and their 95% confidence intervals. RESULT: We enrolled 24,056 adults in four groups (randomized-Fakhravac: 824, randomized-BBIBP-Corv2: 832; Non-randomized-Fakhravac: 19,429, Non-randomized-BBIBP-Corv2: 2971). All observed local and systemic adverse reactions were generally self-limited and resolved completely. We observed similar Serious Adverse Event (SAE) rates in the BBIBP-Corv2 (2.57, 95% CI 1.33-4.49) and Fakhravac (2.25, 95% CI 1.72-2.89) groups; none of which were related to the vaccines received. We recorded 9815 Medically Attendant Adverse Events (MAAE), 736 of which were categorized as somehow related. The rate of related MAAE in the Fakhravac was similar to the BBIBP-Corv2 groups (0.31 and 0.26 per 1000 person-day) in the randomized and considerably higher (0.24 and 0.07 per 1000 person-day) in the non-randomized arms. We observed 129 (35% of the 365 required by target sample size) events of PCR + symptomatic Covid-19 during four months of active follow-up in the randomized arm, demonstrating that those receiving the Fakhravac vaccine were significantly less likely (HR = 0.69; 95% CI 0.49-0.98) to be diagnosed with PCR + symptomatic Covid-19 compared with those receiving BBIBP-Corv2 vaccine. After adjusting for type I error using the O'Brien Fleming method, the Fakhravac vaccine was non-inferior to the BBIBP-Corv2 (assuming a 10% non-inferiority margin to the reported 72.8% BBIBP-Corv2 vaccine efficacy; HR < 1.35) (One-way test: HR = 0.66; 99.8% CI 0.38-1.15). In the non-randomized arm, the results were inconclusive (HR = 1.23; 95% CI 0.96-1.61). We observed 5 cases of hospitalized Covid-19 in the randomized arm, none of which occurred in the Fakhravac vaccine group. Those receiving the Fakhravac vaccine were four times less likely to go to the hospital because of a Covid-19 diagnosis (HR = 0.24; 95% CI 0.10-0.60). The vaccine efficacy of the Fakhravac vaccine is estimated to be 81.5% (95% CI 81-82.4%). CONCLUSION: Fakhravac inactivated SARS-CoV-2 vaccine has comparable safety and efficacy to the BBIBP-Corv2 vaccine. Trial registration This study was registered with the Iranian Registry of Clinical Trials ( www.irct.ir : IRCT20210206050259N3).

Phase II randomized, double blind, placebo controlled, clinical trial of safety and immunogenicity of an inactivated SARS-CoV-2 vaccine FAKHRAVAC in adults aged 18-70 years.

BACKGROUND: The FAKHRAVAC®, an inactivated SARS-CoV-2 vaccine, was assessed for safety and immunogenicity in a phase II trial. METHODS: We did a phase II, single-centered, randomized, double-blind, placebo-controlled clinical trial of the FAKHRAVAC inactivated SARS-CoV-2 vaccine on adults aged 18 to 70. The two parallel groups received two intramuscular injections of either a 10-µg vaccine or a placebo at 2-week intervals. The participants' immunogenicity responses and the occurrence of solicited and unsolicited adverse events were compared over the study period of up to 6 months. Immunogenicity outcomes include serum neutralizing antibody activity and specific IgG antibody levels. RESULTS: Five hundred eligible participants were randomly (1:1) assigned to vaccine or placebo groups. The median age of the participants was 36 years, and 75% were male. The most frequent local adverse reaction was tenderness (21.29% after the first dose and 8.52% after the second dose), and the most frequent systemic adverse reaction was headache (11.24% after the first dose and 8.94% after the second dose). Neutralizing antibody titers two and four weeks after the second injection in the vaccine group showed about 3 and 6 times increase compared to the placebo group (GMR = 2.69, 95% CI 2.32-3.12, N:309) and (GMR = 5.51, 95% CI 3.94-8.35, N:285). A four-fold increase in the neutralizing antibody titer was seen in 69.6% and 73.4% of the participants in the vaccine group two and four weeks after the second dose, respectively. Specific ELIZA antibody response against a combination of S1 and RBD antigens 4 weeks after the second injection increased more than three times in the vaccine compared to the placebo group (GMR = 3.34, 95% CI 2.5-4.47, N:142). CONCLUSIONS: FAKHRAVAC® is safe and induces a significant humoral immune response to the SARS-CoV-2 virus at 10-µg antigen dose in adults aged 18-70. A phase III trial is needed to assess the clinical efficacy. TRIAL REGISTRATION: Trial Registry Number: Ref., IRCT20210206050259N2 ( http://irct.ir ; registered on 08/06/2021).

Comparative Drug Solubility Studies Using Shake-Flask Versus a Laser-Based Robotic Method.

Drug solubility is of central importance to the pharmaceutical sciences, but reported values often show discrepancies. Various factors have been discussed in the literature to account for such differences, but the influence of manual testing in comparison to a robotic system has not been studied adequately before. In this study, four expert researchers were asked to measure the solubility of four drugs with various solubility behaviors (i.e., paracetamol, mesalazine, lamotrigine, and ketoconazole) in the same laboratory with the same instruments, method, and material sources and repeated their measurements after a time interval. In addition, the same solubility data were determined using an automated laser-based setup. The results suggest that manual testing leads to a handling influence on measured solubility values, and the results were discussed in more detail as compared to the automated laser-based system. Within the framework of unavoidable uncertainties of solubility testing, it is a possibility to combine minimal experimental testing that is preferably automated with mathematical modeling. That is a practical suggestion to support future pharmaceutical development in a more efficient way.

Improving the robustness and accuracy of biomedical language models through adversarial training.

Deep transformer neural network models have improved the predictive accuracy of intelligent text processing systems in the biomedical domain. They have obtained state-of-the-art performance scores on a wide variety of biomedical and clinical Natural Language Processing (NLP) benchmarks. However, the robustness and reliability of these models has been less explored so far. Neural NLP models can be easily fooled by adversarial samples, i.e. minor changes to input that preserve the meaning and understandability of the text but force the NLP system to make erroneous decisions. This raises serious concerns about the security and trust-worthiness of biomedical NLP systems, especially when they are intended to be deployed in real-world use cases. We investigated the robustness of several transformer neural language models, i.e. BioBERT, SciBERT, BioMed-RoBERTa, and Bio-ClinicalBERT, on a wide range of biomedical and clinical text processing tasks. We implemented various adversarial attack methods to test the NLP systems in different attack scenarios. Experimental results showed that the biomedical NLP models are sensitive to adversarial samples; their performance dropped in average by 21 and 18.9 absolute percent on character-level and word-level adversarial noise, respectively, on Micro-F1, Pearson Correlation, and Accuracy measures. Conducting extensive adversarial training experiments, we fine-tuned the NLP models on a mixture of clean samples and adversarial inputs. Results showed that adversarial training is an effective defense mechanism against adversarial noise; the models' robustness improved in average by 11.3 absolute percent. In addition, the models' performance on clean data increased in average by 2.4 absolute percent, demonstrating that adversarial training can boost generalization abilities of biomedical NLP systems. This study takes an important step towards revealing vulnerabilities of deep neural language models in biomedical NLP applications. It also provides practical and effective strategies to develop secure, trust-worthy, and accurate intelligent text processing systems in the biomedical domain.

Summarization of biomedical articles using domain-specific word embeddings and graph ranking.

Text summarization tools can help biomedical researchers and clinicians reduce the time and effort needed for acquiring important information from numerous documents. It has been shown that the input text can be modeled as a graph, and important sentences can be selected by identifying central nodes within the graph. However, the effective representation of documents, quantifying the relatedness of sentences, and selecting the most informative sentences are main challenges that need to be addressed in graph-based summarization. In this paper, we address these challenges in the context of biomedical text summarization. We evaluate the efficacy of a graph-based summarizer using different types of context-free and contextualized embeddings. The word representations are produced by pre-training neural language models on large corpora of biomedical texts. The summarizer models the input text as a graph in which the strength of relations between sentences is measured using the domain specific vector representations. We also assess the usefulness of different graph ranking techniques in the sentence selection step of our summarization method. Using the common Recall-Oriented Understudy for Gisting Evaluation (ROUGE) metrics, we evaluate the performance of our summarizer against various comparison methods. The results show that when the summarizer utilizes proper combinations of context-free and contextualized embeddings, along with an effective ranking method, it can outperform the other methods. We demonstrate that the best settings of our graph-based summarizer can efficiently improve the informative content of summaries and decrease the redundancy.

The Effect of Educational Intervention Based on PRECEDE Model on Promoting Skin Cancer Preventive Behaviors in High School Students.

School-based education programs can be an effective way of educating adolescents about the dangers of exposure to sunlight and about preventive measures against this exposure and its relation to skin cancer. The aim of this study is to survey the effect of educational intervention based on the PRECEDE model on promoting skin cancer preventive behaviors in high school students of Fasa City, Fars Province, Iran. In this quasi-experimental study, 300 students (150 in experimental group and 150 in control group) in Fasa City, Fars Province, Iran, were selected in 2016-2017. The educational intervention for the experimental group consisted of six training sessions. A questionnaire consisting of demographic information, PRECEDE constructs (knowledge, attitude, self-efficacy, enabling factors, and social support), was used to measure skin cancer preventive behaviors before and 4 months after the intervention. Data were analyzed using SPSS 22 and paired t test, independent t test, and chi-square test at a significance level of p < 0.05. The mean age of the students was 16.05 ± 1.76 years in the experimental group and 16.20 ± 1.71 years in the control group. Four months after the intervention, the experimental group showed a significant increase in the knowledge, attitude, self-efficacy, enabling factors, social support, and skin cancer preventive behaviors compared to the control group. This study showed the effectiveness of the intervention based on the PRECEDE constructs in adoption of skin cancer preventive behaviors in 4 months post-intervention in students. Hence, this model can act as a framework for designing and implementing educational intervention for the prevention of skin cancer.

CIBS: A biomedical text summarizer using topic-based sentence clustering.

Automatic text summarizers can reduce the time required to read lengthy text documents by extracting the most important parts. Multi-document summarizers should produce a summary that covers the main topics of multiple related input texts to diminish the extent of redundant information. In this paper, we propose a novel summarization method named Clustering and Itemset mining based Biomedical Summarizer (CIBS). The summarizer extracts biomedical concepts from the input documents and employs an itemset mining algorithm to discover main topics. Then, it applies a clustering algorithm to put the sentences into clusters such that those in the same cluster share similar topics. Selecting sentences from all the clusters, the summarizer can produce a summary that covers a wide range of topics of the input text. Using the Recall-Oriented Understudy for Gisting Evaluation (ROUGE) toolkit, we evaluate the performance of the CIBS method against four summarizers including a state-of-the-art method. The results show that the CIBS method can improve the performance of single- and multi-document biomedical text summarization. It is shown that the topic-based sentence clustering approach can be effectively used to increase the informative content of summaries, as well as to decrease the redundant information.

Serum Levels of Indoxyl Sulfate and P-cresol in Type II Diabetic Patients With and Without Nephropathy.

INTRODUCTION: Indoxyl sulfate (IS) and para-cresol (p-cresol) are uremic toxins with high protein bonding index that accumulate in the body with decreasing kidney function. The main purpose of the current investigation was to compare the concentration of p-cresol and IS in serum of the type II diabetic individuals with and without nephropathy. METHODS: Fifty-five patients with type II diabetes mellitus were divided into two groups: case and control. The case group consisted of 26 diabetic patients with nephropathy (proteinuria and serum creatinine below 1.5 mg/dL) without any other kidney diseases. The control group included 29 patients without diabetic nephropathy. Patients with advanced heart disease, cerebrovascular accident and other inflammatory or infectious diseases were excluded. Five mL of venous blood was taken from each patient in the morning fasting state. Then other laboratory tests including serum uric acid and creatinine levels, serum urea nitrogen, lipids and glucose were measured by standard methods. P-Cresol and IS levels were measured by the spectrofluorimetric method after extraction. We also filled out a checklist with information regarding the duration of their disease, medication history (oral or injectable), and other demographic information. There were no significant differences between the two groups regarding the investigated factors Results. There were no significant difference among the investigated factors between the two groups (P > .05) except for the serum creatinine, proteinuria and estimated glomerular filtration rate, where the mean values of cases were considerably higher than those of the controls. Serum IS and p-cresol levels were also significantly higher in the case group (P < .05). CONCLUSION: According to the findings, it seems that IS, and p-cresol may play a role in the development of diabetic nephropathy and other complications of diabetes mellitus.  DOI: 10.52547/ijkd.7266.

ThoughtSource: A central hub for large language model reasoning data.

Large language models (LLMs) such as GPT-4 have recently demonstrated impressive results across a wide range of tasks. LLMs are still limited, however, in that they frequently fail at complex reasoning, their reasoning processes are opaque, they are prone to 'hallucinate' facts, and there are concerns about their underlying biases. Letting models verbalize reasoning steps as natural language, a technique known as chain-of-thought prompting, has recently been proposed as a way to address some of these issues. Here we present ThoughtSource, a meta-dataset and software library for chain-of-thought (CoT) reasoning. The goal of ThoughtSource is to improve future artificial intelligence systems by facilitating qualitative understanding of CoTs, enabling empirical evaluations, and providing training data. This first release of ThoughtSource integrates seven scientific/medical, three general-domain and five math word question answering datasets.

Safety and immunogenicity of an inactivated SARS-CoV-2 vaccine (FAKHRAVAC®) in healthy adults aged 18-55 years: Randomized, double-blind, placebo-controlled, phase I clinical trial.

Background: The FAKHRAVAC®, an inactivated SARS-CoV-2 vaccine, was assessed for safety and immunogenicity. Methods and findings: In this double-blind, placebo-controlled, phase I trial, we randomly assigned 135 healthy adults between 18 and 55 to receive vaccine strengths of 5 or 10 µg/dose or placebo (adjuvant only) in 0-14 or 0-21 schedules. This trial was conducted in a single center in a community setting. The safety outcomes in this study were reactogenicity, local and systemic adverse reactions, abnormal laboratory findings, and Medically Attended Adverse Events (MAAE). Immunogenicity outcomes include serum neutralizing antibody activity and specific IgG antibody levels.The most frequent local adverse reaction was tenderness (28.9%), and the most frequent systemic adverse reaction was headache (9.6%). All adverse reactions were mild, occurred at a similar incidence in all six groups, and were resolved within a few days. In the 10-µg/dose vaccine group, the geometric mean ratio for neutralizing antibody titers at two weeks after the second injection compared to the placebo group was 9.03 (95% CI: 3.89-20.95) in the 0-14 schedule and 11.77 (95% CI: 2.77-49.94) in the 0-21 schedule. The corresponding figures for the 5-µg/dose group were 2.74 (1.2-6.28) and 5.2 (1.63-16.55). The highest seroconversion rate (four-fold increase) was related to the 10-µg/dose group (71% and 67% in the 0-14 and 0-21 schedules, respectively). Conclusions: FAKHRAVAC® is safe and induces a strong humoral immune response to the SARS-CoV-2 virus at 10-µg/dose vaccine strength in adults aged 18-55. This vaccine strength was used for further assessment in the phase II trial.Trial registrationThis study is registered with https://www.irct.ir; IRCT20210206050259N1.

Post-COVID-19 hematologic complications: a systematic review.

INTRODUCTION: COVID-19 crisis continues around the world. Some patients developed complications after the disease, which have been reported in limited studies. The aim of this study is to comprehensively assess the post-COVID hematologic complications in patients. AREAS COVERED: We searched PubMed, Scopus, and Google Scholar between January 2020 and August 2021 using related keywords. Evaluation of the article was performed by two independent researchers. The extracted data included the number of patients, age, type of hematological complication, duration of follow-up, response to treatment and prognosis. EXPERT OPINION: Sixty-five articles reported post-COVID hematologic complications. The most frequent hematologic complication in COVID-19 patients is thromboembolic events, which often occur in two forms: deep vein thrombosis (DVT) and pulmonary embolism (PE). In a group of patients after the diagnosis of COVID-19, a significant decrease in platelets was observed, which was attributed to the ITP induced by COVID-19. Hemolytic anemia and aplastic anemia have also been reported rarely in patients. Finally, post-COVID hematologic complications appear to go beyond thromboembolic events. Although these complications have rarely been reported, searching for methods to identify susceptible patients and prevent these complications could be the subject of future research.

A combination of amino-functionalized fibrous silica (KCC-1-NH2)/effectively and efficiently oxidized graphene oxide (EEGO) nanocomposite for dispersive solid-phase extraction, pre-concentration and fluorescence determination of total para-cresol in plasma samples of chronic kidney disease patients.

A novel amino-functionalized fibrous silica (KCC-1-NH2) and effectively and efficiently oxidized graphene oxide (EEGO) nanocomposite has been successfully synthesized. This nanocomposite was applied as a new sorbent in the dispersive solid-phase extraction (-SPE) to the preconcentration of total p-cresyl sulfate (pCS) in human plasma samples. The morphology and basic structure of the proposed nanocomposite were investigated through different techniques including field emission scanning electron microscopy (FESEM), energy dispersive X-ray (EDX), transmission electron microscopy (TEM), Fourier transform infrared (FTIR), and dynamic light scattering (DLS)/zeta potential techniques. The influence of different factors on the extraction efficiency, including the amount of sorbent, sample pH, extraction time, elution solvents and their volume, and desorption time were also investigated. The developed fluorescence-based method offers a linear dynamic range from 0.02 to 6 µg/mL with an acceptable correlation coefficient (R2 = 0.9982) and recovery (80%). The limit of quantification (LOQ) and limit of detection (LOD) were found to be 0.043 and 0.013 µg/mL, respectively. Plasma samples of five chronic kidney disease (CKD) patients were also analyzed and measured pCS concentrations were ranged from 16 to 41 µg/mL. The applicability of the method was successfully tested for the extraction and quantification of pCS from spiked and patients' plasma samples.

Comparison of the Safety and Immunogenicity of FAKHRAVAC and BBIBP-CorV Vaccines when Administrated as Booster Dose: A Parallel Two Arms, Randomized, Double Blind Clinical Trial.

Purpose: This study was completed to assess the immunogenicity and safety of the FAKHRAVAC and BBIBP-CorV vaccines as a booster dose in the population with a history of receiving two doses of BBIBP-CorV vaccine. Methods: In this double-blind, parallel clinical trial, we randomly assigned healthy adults with a history of receiving two doses of the BBIBP-CorV vaccine, who then received either the FAKHRAVAC or BBIBP-CorV vaccine as a booster dose. The trial is registered in the Iranian Registry of Clinical Trial document depository (Code: IRCT20210206050259N4). Results: The outcomes that were monitored in this study were serum neutralizing antibody (Nab) activity, immunoglobulin G (IgG) level, local and systemic adverse reactions, serious adverse events, suspected unexpected serious adverse reactions, and medically attended adverse events. After administering vaccines to 435 participants, the most frequent local and systemic adverse reactions were tenderness and nausea in 23.7% and 1.4% of cases, respectively. All adverse events were mild, occurred at a similar incidence in the two groups, and were resolved within a few days. Conclusions: On the 14th day after the booster dose injection, the seroconversion rate (i.e., four-fold increase) of Nabs for seronegative participants were 87% and 84.6% in the FAKHRAVAC® and BBIBP-CorV groups, respectively. This study shows that the FAKHRAVAC® vaccine, as a booster dose, has a similar function to the BBIBP-CorV vaccine in terms of increasing the titer of virus-neutralizing antibodies, the amount of specific antibodies, and safety.

Explaining Black-Box Models for Biomedical Text Classification.

In this paper, we propose a novel method named Biomedical Confident Itemsets Explanation (BioCIE), aiming at post-hoc explanation of black-box machine learning models for biomedical text classification. Using sources of domain knowledge and a confident itemset mining method, BioCIE discretizes the decision space of a black-box into smaller subspaces and extracts semantic relationships between the input text and class labels in different subspaces. Confident itemsets discover how biomedical concepts are related to class labels in the black-box's decision space. BioCIE uses the itemsets to approximate the black-box's behavior for individual predictions. Optimizing fidelity, interpretability, and coverage measures, BioCIE produces class-wise explanations that represent decision boundaries of the black-box. Results of evaluations on various biomedical text classification tasks and black-box models demonstrated that BioCIE can outperform perturbation-based and decision set methods in terms of producing concise, accurate, and interpretable explanations. BioCIE improved the fidelity of instance-wise and class-wise explanations by 11.6% and 7.5%, respectively. It also improved the interpretability of explanations by 8%. BioCIE can be effectively used to explain how a black-box biomedical text classification model semantically relates input texts to class labels. The source code and supplementary material are available at https://github.com/mmoradi-iut/BioCIE.

Simple Determination of p-Cresol in Plasma Samples Using Fluorescence Spectroscopy Technique.

The development of simple, fast, cheap and reliable analytical methods for tracing biological indicators is demanded through clinical investigations. Herein, we developed, for the first time, a cheap and specific method for the extraction and quantification of p-cresol (pC) in real plasma samples of chronic kidney disease (CKD). Plasma samples were prepared by hydrolyzing in an acidic medium to convert pCS (p-cresol sulfate) and p-Cresol glucuronide (pCG) to pC. Next, proteins of plasma samples were precipitated and then pC was extracted by acetonitrile (ACN) and saturated NaCl (as salting-out agent). Finally, fluorescence emissions were measured at λex/λem = 280/310 nm. The specificity of the method was checked by testing various possible interfering agents. The obtained results revealed a specific determination of pC. Under optimal conditions, a linear range was detected from 0.5 to 30 µg/mL of pC with a lower limit of detection (LLOQ) of 0.5 µg/mL. The reliability of the method was checked by calculating the repeatability, selectivity, and accuracy of the developed method for pC determination in plasma samples. The application of the developed method was investigated for the detection of pC in a number of CKD patients. Due to the simplicity and selectivity, the developed method could be applied for routine analysis of pC concentrations in the plasma samples of CKD patients. In addition, the developed method showed great potential for developing a point-of-care testing (POCT) device.

Comprehensive Review on Neuro-COVID-19 Pathophysiology and Clinical Consequences.

Aside from the respiratory distress as the predominant clinical presentation of SARS-CoV-2 infection, various neurological complications have been reported with the infection during the ongoing pandemic, some of which cause serious morbidity and mortality. Herein, we gather the latest anatomical evidence of the virus's presence within the central nervous system. We then delve into the possible SARS-CoV-2 entry routes into the neurological tissues, with the hematogenous and the neuronal routes as the two utmost passage routes into the nervous system. We then give a comprehensive review of the neurological manifestations of the SARS-CoV-2 invasion in both the central and peripheral nervous system and its underlying pathophysiology via investigating large studies in the field and case reports in cases of study scarcity.

Deep contextualized embeddings for quantifying the informative content in biomedical text summarization.

BACKGROUND AND OBJECTIVE: Capturing the context of text is a challenging task in biomedical text summarization. The objective of this research is to show how contextualized embeddings produced by a deep bidirectional language model can be utilized to quantify the informative content of sentences in biomedical text summarization. METHODS: We propose a novel summarization method that utilizes contextualized embeddings generated by the Bidirectional Encoder Representations from Transformers (BERT) model, a deep learning model that recently demonstrated state-of-the-art results in several natural language processing tasks. We combine different versions of BERT with a clustering method to identify the most relevant and informative sentences of input documents. Using the ROUGE toolkit, we evaluate the summarizer against several methods previously described in literature. RESULTS: The summarizer obtains state-of-the-art results and significantly improves the performance of biomedical text summarization in comparison to a set of domain-specific and domain-independent methods. The largest language model not specifically pretrained on biomedical text outperformed other models. However, among language models of the same size, the one further pretrained on biomedical text obtained best results. CONCLUSIONS: We demonstrate that a hybrid system combining a deep bidirectional language model and a clustering method yields state-of-the-art results without requiring labor-intensive creation of annotated features or knowledge bases or computationally demanding domain-specific pretraining. This study provides a starting point towards investigating deep contextualized language models for biomedical text summarization.

Different approaches for identifying important concepts in probabilistic biomedical text summarization.

Automatic text summarization tools help users in the biomedical domain to acquire their intended information from various textual resources more efficiently. Some of biomedical text summarization systems put the basis of their sentence selection approach on the frequency of concepts extracted from the input text. However, it seems that exploring other measures rather than the raw frequency for identifying valuable contents within an input document, or considering correlations existing between concepts, may be more useful for this type of summarization. In this paper, we describe a Bayesian summarization method for biomedical text documents. The Bayesian summarizer initially maps the input text to the Unified Medical Language System (UMLS) concepts; then it selects the important ones to be used as classification features. We introduce six different feature selection approaches to identify the most important concepts of the text and select the most informative contents according to the distribution of these concepts. We show that with the use of an appropriate feature selection approach, the Bayesian summarizer can improve the performance of biomedical summarization. Using the Recall-Oriented Understudy for Gisting Evaluation (ROUGE) toolkit, we perform extensive evaluations on a corpus of scientific papers in the biomedical domain. The results show that when the Bayesian summarizer utilizes the feature selection methods that do not use the raw frequency, it can outperform the biomedical summarizers that rely on the frequency of concepts, domain-independent and baseline methods.

Quantifying the informativeness for biomedical literature summarization: An itemset mining method.

OBJECTIVE: Automatic text summarization tools can help users in the biomedical domain to access information efficiently from a large volume of scientific literature and other sources of text documents. In this paper, we propose a summarization method that combines itemset mining and domain knowledge to construct a concept-based model and to extract the main subtopics from an input document. Our summarizer quantifies the informativeness of each sentence using the support values of itemsets appearing in the sentence. METHODS: To address the concept-level analysis of text, our method initially maps the original document to biomedical concepts using the Unified Medical Language System (UMLS). Then, it discovers the essential subtopics of the text using a data mining technique, namely itemset mining, and constructs the summarization model. The employed itemset mining algorithm extracts a set of frequent itemsets containing correlated and recurrent concepts of the input document. The summarizer selects the most related and informative sentences and generates the final summary. RESULTS: We evaluate the performance of our itemset-based summarizer using the Recall-Oriented Understudy for Gisting Evaluation (ROUGE) metrics, performing a set of experiments. We compare the proposed method with GraphSum, TexLexAn, SweSum, SUMMA, AutoSummarize, the term-based version of the itemset-based summarizer, and two baselines. The results show that the itemset-based summarizer performs better than the compared methods. The itemset-based summarizer achieves the best scores for all the assessed ROUGE metrics (R-1: 0.7583, R-2: 0.3381, R-W-1.2: 0.0934, and R-SU4: 0.3889). We also perform a set of preliminary experiments to specify the best value for the minimum support threshold used in the itemset mining algorithm. The results demonstrate that the value of this threshold directly affects the accuracy of the summarization model, such that a significant decrease can be observed in the performance of summarization due to assigning extreme thresholds. CONCLUSION: Compared to the statistical, similarity, and word frequency methods, the proposed method demonstrates that the summarization model obtained from the concept extraction and itemset mining provides the summarizer with an effective metric for measuring the informative content of sentences. This can lead to an improvement in the performance of biomedical literature summarization.