Large-scale language analysis of peer review reports.
Ontology highlight
ABSTRACT:
SUBMITTER: Buljan I
PROVIDER: S-EPMC7390598 | biostudies-literature | 2020 Jul
REPOSITORIES: biostudies-literature
Publications
Large-scale language analysis of peer review reports.
Buljan Ivan I Garcia-Costa Daniel D Grimaldo Francisco F Squazzoni Flaminio F Marušić Ana A
eLife 20200717
Peer review is often criticized for being flawed, subjective and biased, but research into peer review has been hindered by a lack of access to peer review reports. Here we report the results of a study in which text-analysis software was used to determine the linguistic characteristics of 472,449 peer review reports. A range of characteristics (including analytical tone, authenticity, clout, three measures of sentiment, and morality) were studied as a function of reviewer recommendation, area o ...[more]
Similar Datasets
Project description:Large volumes of data are continuously generated from clinical notes and diagnostic studies catalogued in electronic health records (EHRs). Echocardiography is one of the most commonly ordered diagnostic tests in cardiology. This study sought to explore the feasibility and reliability of using natural language processing (NLP) for large-scale and targeted extraction of multiple data elements from echocardiography reports. An NLP tool, EchoInfer, was developed to automatically extract data pertaining to cardiovascular structure and function from heterogeneously formatted echocardiographic data sources. EchoInfer was applied to echocardiography reports (2004 to 2013) available from 3 different on-going clinical research projects. EchoInfer analyzed 15,116 echocardiography reports from 1684 patients, and extracted 59 quantitative and 21 qualitative data elements per report. EchoInfer achieved a precision of 94.06%, a recall of 92.21%, and an F1-score of 93.12% across all 80 data elements in 50 reports. Physician review of 400 reports demonstrated that EchoInfer achieved a recall of 92-99.9% and a precision of >97% in four data elements, including three quantitative and one qualitative data element. Failure of EchoInfer to correctly identify or reject reported parameters was primarily related to non-standardized reporting of echocardiography data. EchoInfer provides a powerful and reliable NLP-based approach for the large-scale, targeted extraction of information from heterogeneous data sources. The use of EchoInfer may have implications for the clinical management and research analysis of patients undergoing echocardiographic evaluation.
| S-EPMC4849652 | biostudies-literature
Project description:ObjectivesTo examine whether the gender of applicants and peer reviewers and other factors influence peer review of grant proposals submitted to a national funding agency.SettingSwiss National Science Foundation (SNSF).DesignCross-sectional analysis of peer review reports submitted from 2009 to 2016 using linear mixed effects regression models adjusted for research topic, applicant's age, nationality, affiliation and calendar period.ParticipantsExternal peer reviewers.Primary outcome measureOverall score on a scale from 1 (worst) to 6 (best).ResultsAnalyses included 38 250 reports on 12 294 grant applications from medicine, architecture, biology, chemistry, economics, engineering, geology, history, linguistics, mathematics, physics, psychology and sociology submitted by 26 829 unique peer reviewers. In univariable analysis, male applicants received more favourable evaluation scores than female applicants (+0.18 points; 95% CI 0.14 to 0.23), and male reviewers awarded higher scores than female reviewers (+0.11; 95% CI 0.08 to 0.15). Applicant-nominated reviewers awarded higher scores than reviewers nominated by the SNSF (+0.53; 95% CI 0.50 to 0.56), and reviewers from outside of Switzerland more favourable scores than reviewers affiliated with Swiss institutions (+0.53; 95% CI 0.49 to 0.56). In multivariable analysis, differences between male and female applicants were attenuated (+0.08; 95% CI 0.04 to 0.13) whereas results changed little for source of nomination and affiliation of reviewers. The gender difference increased after September 2011, when new evaluation forms were introduced (p=0.033 from test of interaction).ConclusionsPeer review of grant applications at SNSF might be prone to biases stemming from different applicant and reviewer characteristics. The SNSF abandoned the nomination of peer reviewers by applicants. The new form introduced in 2011 may inadvertently have given more emphasis to the applicant's track record. We encourage other funders to conduct similar studies, in order to improve the evidence base for rational and fair research funding.
| S-EPMC7440717 | biostudies-literature
Project description:ObjectivesTo assess the extent to which peer reviewers and journals editors address study funding and authors' conflicts of interests (COI). Also, we aimed to assess the extent to which peer reviewers and journals editors reported and commented on their own or each other's COI.Study design and methodsWe conducted a systematic survey of original studies published in open access peer reviewed journals that publish their peer review reports. Using REDCap, we collected data in duplicate and independently from journals' websites and articles' peer review reports.ResultsWe included a sample of original studies (N = 144) and a second one of randomized clinical trials (N = 115) RCTs. In both samples, and for the majority of studies, reviewers reported absence of COI (70% and 66%), while substantive percentages of reviewers did not report on COI (28% and 30%) and only small percentages reported any COI (2% and 4%). For both samples, none of the editors whose names were publicly posted reported on COI. The percentages of peer reviewers commenting on the study funding, authors' COI, editors' COI, or their own COI ranged between 0 and 2% in either one of the two samples. 25% and 7% of editors respectively in the two samples commented on study funding, while none commented on authors' COI, peer reviewers' COI, or their own COI. The percentages of authors commenting in their response letters on the study funding, peer reviewers' COI, editors' COI, or their own COI ranged between 0 and 3% in either one of the two samples.ConclusionThe percentages of peer reviewers and journals editors who addressed study funding and authors' COI and were extremely low. In addition, peer reviewers and journal editors rarely reported their own COI, or commented on their own or on each other's COI.
| S-EPMC10249818 | biostudies-literature
Project description:IntroductionRadiological imaging is one of the most frequently performed diagnostic tests worldwide. The free-text contained in radiology reports is currently only rarely used for secondary use purposes, including research and predictive analysis. However, this data might be made available by means of information extraction (IE), based on natural language processing (NLP). Recently, a new approach to NLP, large language models (LLMs), has gained momentum and continues to improve performance of IE-related tasks. The objective of this scoping review is to show the state of research regarding IE from free-text radiology reports based on LLMs, to investigate applied methods and to guide future research by showing open challenges and limitations of current approaches. To our knowledge, no systematic or scoping review of IE from radiology reports based on LLMs has been published. Existing publications are outdated and do not comprise LLM-based methods.Methods and analysisThis protocol is designed based on the JBI Manual for Evidence Synthesis, chapter 11.2: 'Development of a scoping review protocol'. Inclusion criteria and a search strategy comprising four databases (PubMed, IEEE Xplore, Web of Science Core Collection and ACM Digital Library) are defined. Furthermore, we describe the screening process, data charting, analysis and presentation of extracted data.Ethics and disseminationThis protocol describes the methodology of a scoping literature review and does not comprise research on or with humans, animals or their data. Therefore, no ethical approval is required. After the publication of this protocol and the conduct of the review, its results are going to be published in an open access journal dedicated to biomedical informatics/digital health.
| S-EPMC10729196 | biostudies-literature
Project description:BackgroundThe Patient-Centered Outcomes Research Institute (PCORI) is obligated to peer review and to post publicly "Final Research Reports" of all funded projects. PCORI peer review emphasizes adherence to PCORI's Methodology Standards and principles of ethical scientific communication. During the peer review process, reviewers and editors seek to ensure that results are presented objectively and interpreted appropriately, e.g., free of spin.MethodsTwo independent raters assessed PCORI peer review feedback sent to authors. We calculated the proportion of reports in which spin was identified during peer review, and the types of spin identified. We included reports submitted by April 2018 with at least one associated journal article. The same raters then assessed whether authors addressed reviewers' comments about spin. The raters also assessed whether spin identified during PCORI peer review was present in related journal articles.ResultsWe included 64 PCORI-funded projects. Peer reviewers or editors identified spin in 55/64 (86%) submitted research reports. Types of spin included reporting bias (46/55; 84%), inappropriate interpretation (40/55; 73%), inappropriate extrapolation of results (15/55; 27%), and inappropriate attribution of causality (5/55; 9%). Authors addressed comments about spin related to 47/55 (85%) of the reports. Of 110 associated journal articles, PCORI comments about spin were potentially applicable to 44/110 (40%) articles, of which 27/44 (61%) contained the same spin that was identified in the PCORI research report. The proportion of articles with spin was similar for articles accepted before and after PCORI peer review (63% vs 58%).DiscussionJust as spin is common in journal articles and press releases, we found that most reports submitted to PCORI included spin. While most spin was mitigated during the funder's peer review process, we found no evidence that review of PCORI reports influenced spin in journal articles. Funders could explore interventions aimed at reducing spin in published articles of studies they support.
| S-EPMC8638354 | biostudies-literature
Project description:Reviewers do not only help editors to screen manuscripts for publication in academic journals; they also serve to increase the rigor and value of manuscripts by constructive feedback. However, measuring this developmental function of peer review is difficult as it requires fine-grained data on reports and journals without any optimal benchmark. To fill this gap, we adapted a recently proposed quality assessment tool and tested it on a sample of 1.3 million reports submitted to 740 Elsevier journals in 2018-2020. Results showed that the developmental standards of peer review are shared across areas of research, yet with remarkable differences. Reports submitted to social science and economics journals show the highest developmental standards. Reports from junior reviewers, women and reviewers from Western Europe are generally more developmental than those from senior, men and reviewers working in academic institutions outside Western regions. Our findings suggest that increasing the standards of peer review at journals requires effort to assess interventions and measure practices with context-specific and multi-dimensional frameworks.
| S-EPMC9186327 | biostudies-literature
Project description:PurposeTo determine if fine-tuned large language models (LLMs) can generate accurate, personalized impressions for whole-body PET reports.Materials and methodsTwelve language models were trained on a corpus of PET reports using the teacher-forcing algorithm, with the report findings as input and the clinical impressions as reference. An extra input token encodes the reading physician's identity, allowing models to learn physician-specific reporting styles. Our corpus comprised 37,370 retrospective PET reports collected from our institution between 2010 and 2022. To identify the best LLM, 30 evaluation metrics were benchmarked against quality scores from two nuclear medicine (NM) physicians, with the most aligned metrics selecting the model for expert evaluation. In a subset of data, model-generated impressions and original clinical impressions were assessed by three NM physicians according to 6 quality dimensions (3-point scale) and an overall utility score (5-point scale). Each physician reviewed 12 of their own reports and 12 reports from other physicians. Bootstrap resampling was used for statistical analysis.ResultsOf all evaluation metrics, domain-adapted BARTScore and PEGASUSScore showed the highest Spearman's ρ correlations (ρ=0.568 and 0.563) with physician preferences. Based on these metrics, the fine-tuned PEGASUS model was selected as the top LLM. When physicians reviewed PEGASUS-generated impressions in their own style, 89% were considered clinically acceptable, with a mean utility score of 4.08 out of 5. Physicians rated these personalized impressions as comparable in overall utility to the impressions dictated by other physicians (4.03, P=0.41).ConclusionPersonalized impressions generated by PEGASUS were clinically useful, highlighting its potential to expedite PET reporting.
| S-EPMC10614982 | biostudies-literature
Project description:As metagenomic studies continue to increase in their number, sequence volume and complexity, the scalability of biological analysis frameworks has become a rate-limiting factor to meaningful data interpretation. To address this issue, we have developed JCVI Metagenomics Reports (METAREP) as an open source tool to query, browse, and compare extremely large volumes of metagenomic annotations. Here we present improvements to this software including the implementation of a dynamic weighting of taxonomic and functional annotation, support for distributed searches, advanced clustering routines, and integration of additional annotation input formats. The utility of these improvements to data interpretation are demonstrated through the application of multiple comparative analysis strategies to shotgun metagenomic data produced by the National Institutes of Health Roadmap for Biomedical Research Human Microbiome Project (HMP) (http://nihroadmap.nih.gov). Specifically, the scalability of the dynamic weighting feature is evaluated and established by its application to the analysis of over 400 million weighted gene annotations derived from 14 billion short reads as predicted by the HMP Unified Metabolic Analysis Network (HUMAnN) pipeline. Further, the capacity of METAREP to facilitate the identification and simultaneous comparison of taxonomic and functional annotations including biological pathway and individual enzyme abundances from hundreds of community samples is demonstrated by providing scenarios that describe how these data can be mined to answer biological questions related to the human microbiome. These strategies provide users with a reference of how to conduct similar large-scale metagenomic analyses using METAREP with their own sequence data, while in this study they reveal insights into the nature and extent of variation in taxonomic and functional profiles across body habitats and individuals. Over one thousand HMP WGS datasets and the latest open source code are available at http://www.jcvi.org/hmp-metarep.
| S-EPMC3374610 | biostudies-literature
Project description:ObjectiveTo investigate the effectiveness of open peer review as a mechanism to improve the reporting of randomised trials published in biomedical journals.DesignRetrospective before and after study.SettingBioMed Central series medical journals.Sample93 primary reports of randomised trials published in BMC-series medical journals in 2012.Main outcome measuresChanges to the reporting of methodological aspects of randomised trials in manuscripts after peer review, based on the CONSORT checklist, corresponding peer reviewer reports, the type of changes requested, and the extent to which authors adhered to these requests.ResultsOf the 93 trial reports, 38% (n=35) did not describe the method of random sequence generation, 54% (n=50) concealment of allocation sequence, 50% (n=46) whether the study was blinded, 34% (n=32) the sample size calculation, 35% (n=33) specification of primary and secondary outcomes, 55% (n=51) results for the primary outcome, and 90% (n=84) details of the trial protocol. The number of changes between manuscript versions was relatively small; most involved adding new information or altering existing information. Most changes requested by peer reviewers had a positive impact on the reporting of the final manuscript--for example, adding or clarifying randomisation and blinding (n=27), sample size (n=15), primary and secondary outcomes (n=16), results for primary or secondary outcomes (n=14), and toning down conclusions to reflect the results (n=27). Some changes requested by peer reviewers, however, had a negative impact, such as adding additional unplanned analyses (n=15).ConclusionPeer reviewers fail to detect important deficiencies in reporting of the methods and results of randomised trials. The number of these changes requested by peer reviewers was relatively small. Although most had a positive impact, some were inappropriate and could have a negative impact on reporting in the final publication.
| S-EPMC4077234 | biostudies-literature
Project description:BACKGROUND:A strong need exists for a validated tool that clearly defines peer review report quality in biomedical research, as it will allow evaluating interventions aimed at improving the peer review process in well-performed trials. We aim to identify and describe existing tools for assessing the quality of peer review reports in biomedical research. METHODS:We conducted a methodological systematic review by searching PubMed, EMBASE (via Ovid) and The Cochrane Methodology Register (via The Cochrane Library) as well as Google® for all reports in English describing a tool for assessing the quality of a peer review report in biomedical research. Data extraction was performed in duplicate using a standardized data extraction form. We extracted information on the structure, development and validation of each tool. We also identified quality components across tools using a systematic multi-step approach and we investigated quality domain similarities among tools by performing hierarchical, complete-linkage clustering analysis. RESULTS:We identified a total number of 24 tools: 23 scales and 1 checklist. Six tools consisted of a single item and 18 had several items ranging from 4 to 26. None of the tools reported a definition of 'quality'. Only 1 tool described the scale development and 10 provided measures of validity and reliability. Five tools were used as an outcome in a randomized controlled trial (RCT). Moreover, we classified the quality components of the 18 tools with more than one item into 9 main quality domains and 11 subdomains. The tools contained from two to seven quality domains. Some domains and subdomains were considered in most tools such as the detailed/thorough (11/18) nature of reviewer's comments. Others were rarely considered, such as whether or not the reviewer made comments on the statistical methods (1/18). CONCLUSION:Several tools are available to assess the quality of peer review reports; however, the development and validation process is questionable and the concepts evaluated by these tools vary widely. The results from this study and from further investigations will inform the development of a new tool for assessing the quality of peer review reports in biomedical research.
| S-EPMC6402095 | biostudies-literature