Dataset Information

Closing the loop: automatically identifying abnormal imaging results in scanned documents.

ABSTRACT:

Objectives

Scanned documents (SDs), while common in electronic health records and potentially rich in clinically relevant information, rarely fit well with clinician workflow. Here, we identify scanned imaging reports requiring follow-up with high recall and practically useful precision.

Materials and methods

We focused on identifying imaging findings for 3 common causes of malpractice claims: (1) potentially malignant breast (mammography) and (2) lung (chest computed tomography [CT]) lesions and (3) long-bone fracture (X-ray) reports. We train our ClinicalBERT-based pipeline on existing typed/dictated reports classified manually or using ICD-10 codes, evaluate using a test set of manually classified SDs, and compare against string-matching (baseline approach).

Results

A total of 393 mammograms, 305 chest CT, and 683 bone X-ray reports were manually reviewed. The string-matching approach had an F1 of 0.667. For mammograms, chest CTs, and bone X-rays, respectively: models trained on manually classified training data and optimized for F1 reached an F1 of 0.900, 0.905, and 0.817, while separate models optimized for recall achieved a recall of 1.000 with precisions of 0.727, 0.518, and 0.275. Models trained on ICD-10-labelled data and optimized for F1 achieved F1 scores of 0.647, 0.830, and 0.643, while those optimized for recall achieved a recall of 1.0 with precisions of 0.407, 0.683, and 0.358.

Discussion

Our pipeline can identify abnormal reports with potentially useful performance and so decrease the manual effort required to screen for abnormal findings that require follow-up.

Conclusion

It is possible to automatically identify clinically significant abnormalities in SDs with high recall and practically useful precision in a generalizable and minimally laborious way.

SUBMITTER: Kumar A

PROVIDER: S-EPMC9714594 | biostudies-literature | 2022 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Closing the loop: automatically identifying abnormal imaging results in scanned documents.

Kumar Akshat A Goodrum Heath H Kim Ashley A Stender Carly C Roberts Kirk K Bernstam Elmer V EV

Journal of the American Medical Informatics Association : JAMIA 20220401 5

<h4>Objectives</h4>Scanned documents (SDs), while common in electronic health records and potentially rich in clinically relevant information, rarely fit well with clinician workflow. Here, we identify scanned imaging reports requiring follow-up with high recall and practically useful precision.<h4>Materials and methods</h4>We focused on identifying imaging findings for 3 common causes of malpractice claims: (1) potentially malignant breast (mammography) and (2) lung (chest computed tomography [ ...[more]

PMID: 35146510

Similar Datasets

Project description:BackgroundCellular events play a central role in the understanding of biological processes and functions, providing insight on both physiological and pathogenesis mechanisms. Automatic extraction of mentions of such events from the literature represents an important contribution to the progress of the biomedical domain, allowing faster updating of existing knowledge. The identification of trigger words indicating an event is a very important step in the event extraction pipeline, since the following task(s) rely on its output. This step presents various complex and unsolved challenges, namely the selection of informative features, the representation of the textual context, and the selection of a specific event type for a trigger word given this context.ResultsWe propose TrigNER, a machine learning-based solution for biomedical event trigger recognition, which takes advantage of Conditional Random Fields (CRFs) with a high-end feature set, including linguistic-based, orthographic, morphological, local context and dependency parsing features. Additionally, a completely configurable algorithm is used to automatically optimize the feature set and training parameters for each event type. Thus, it automatically selects the features that have a positive contribution and automatically optimizes the CRF model order, n-grams sizes, vertex information and maximum hops for dependency parsing features. The final output consists of various CRF models, each one optimized to the linguistic characteristics of each event type.ConclusionsTrigNER was tested in the BioNLP 2009 shared task corpus, achieving a total F-measure of 62.7 and outperforming existing solutions on various event trigger types, namely gene expression, transcription, protein catabolism, phosphorylation and binding. The proposed solution allows researchers to easily apply complex and optimized techniques in the recognition of biomedical event triggers, making its application a simple routine task. We believe this work is an important contribution to the biomedical text mining community, contributing to improved and faster event recognition on scientific articles, and consequent hypothesis generation and knowledge discovery. This solution is freely available as open source at http://bioinformatics.ua.pt/trigner.

Project description:BackgroundAdvances in diabetes technology have been exponential in the last few decades. With evolution in continuous glucose monitoring (CGM) systems and its progressive automation in control of insulin delivery, these advances have changed type 1 diabetes mellitus (T1DM) management. These novel technologies have the potential to improve glycated haemoglobin (HbA1c), reduce hypoglycaemic events, increase time spent in range and improve quality of life (QoL). Our aim was to evaluate the sustained effects in free-living unsupervised conditions of CGM systems (intermittently scanned and real time) and insulin delivery [from multiple daily injections, via sensor-augmented pump therapy and (predictive) low-glucose insulin suspension to hybrid closed-loop systems] on glucose control and QoL in adults and children with T1DM.MethodsWe performed a systematic review of randomized controlled trials (RCTs), using PubMed and the Cochrane library up to 30 May 2019. Inclusion of RCTs was based on type of intervention (comparing glucose-monitoring devices and insulin-delivery devices), population (nonpregnant adults and children with T1DM), follow-up (outpatient setting for at least 8 weeks) and relevant outcomes [HbA1c, time in range (TIR), time in target, time in hypoglycaemia and QoL]. Exclusion of RCTs was based on intervention (exercise, only overnight use). The Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines were used to score the quality of the papers and for the final selection of the articles.ResultsOur search resulted in 214 articles, of which 19 were eligible. Studies on advanced use in adults and children with T1DM reported increased TIR (all 9 studies); decreased time in hypoglycaemia (13 out of 15 studies); lowered HbA1c levels (5 out of 15 studies); improved QoL (10 of 16 studies) and treatment satisfaction (7 studies).ConclusionsRecent technologies have dramatically changed the course of T1DM. They are proving useful in controlling glycaemia in patients with T1DM, without increasing the treatment burden.

Project description:Frontotemporal lobar degeneration comprises a group of disorders characterized by behavioural, executive, language impairment and sometimes features of parkinsonism and motor neuron disease. In 1994 we described an Irish-American family with frontotemporal dementia linked to chromosome 17 associated with extensive tau pathology. We named this disinhibition-dementia-parkinsonism-amyotrophy complex. We subsequently identified mutations in the MAPT gene. Eleven MAPT gene splice site stem loop mutations were identified over time except for 5' splice site of exon 10. We recently identified another Irish family with autosomal dominant early amnesia and behavioural change or parkinsonism associated with the 'missing' +15 mutation at the intronic boundary of exon 10. We performed a clinical, neuropsychological and neuroimaging study on the proband and four siblings, including two affected siblings. We sequenced MAPT and performed segregation analysis. We looked for a biological effect of the tau variant by performing real-time polymerase chain reaction analysis of RNA extracted from human embryonic kidney cells transfected with exon trapping constructs. We found a c.915+15A>C exon 10/intron 10 stem loop mutation in all affected subjects but not in the unaffected. The c.915+15A>C variant caused a shift in tau splicing pattern to a predominantly exon 10+ pattern presumably resulting in predominant 4 repeat tau and little 3 repeat tau. This strongly suggests that the c.915+15A>C variant is a mutation and that it causes frontotemporal dementia linked to chromosome 17 in this pedigree by shifting tau transcription and translation to +4 repeat tau. Tau (MAPT) screening should be considered in families where amnesia or atypical parkinsonism coexists with behavioural disturbance early in the disease process. We describe the final missing stem loop tau mutation predicted 15 years ago. Mutations have now been identified at all predicted sites within the 'stem' when the stem-loop model was first proposed and no mutations have been found within the 'loop' region as expected. Therefore we 'close the tau loop' having 'opened the loop' 21 years ago.

Project description:BackgroundCommunication failures involving test results contribute to issues of patient harm and sentinel events. This article aims to synthesise review evidence, practice insights and patient perspectives addressing problems encountered in the communication of diagnostic test results.MethodsThe rapid review identified ten systematic reviews and four narrative reviews. Five practitioner interviews identified insights into interventions and implementation, and a citizen panel with 15 participants explored the patient viewpoint.ResultsThe rapid review provided support for the role of technology to ensure effective communication; behavioural interventions such as audit and feedback could be effective in changing clinician behaviour; and point-of-care tests (bedside testing) eliminate the communication breakdown problem altogether. The practice interviews highlighted transparency, and clarifying the lines of responsibility as central to improving test result communication. Enabling better information sharing, implementing adequate planning and utilising technology were also identified in the practice interviews as viable strategies to improve test result communication. The citizen panel highlighted technology as critical to improving communication of test results to both health professionals and patients. Patients also highlighted the importance of having different ways of accessing test results, which is particularly pertinent when ensuring suitability for vulnerable populations.ConclusionsThis paper draws together multiple perspectives on the problem of failures in diagnostic test results communication to inform appropriate interventions. Across the three studies, technology was identified as the most feasible option for closing the loop on test result communication. However, the importance of clear, consistent communication and more streamlined processes were also key elements that emerged.Review registrationThe protocol for the rapid review was registered with PROSPERO CRD42018093316 .

Dataset Information

Closing the loop: automatically identifying abnormal imaging results in scanned documents.

Objectives

Materials and methods

Results

Discussion

Conclusion

Publications

Closing the loop: automatically identifying abnormal imaging results in scanned documents.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets