Data Extraction Tasks Research Articles

Porphyry-type mineral prospectivity mapping with imbalanced data via prior geological transfer learning

Mineral prospectivity mapping is crucial for identifying areas with economically valuable minerals. Therefore, several methods based on machine learning have been applied to predict the likelihood of mineral occurrences, especially deep learning (DL), which provides a flexible and precise approach to the use of continuous data. It allows the approximation of predictive variables with probability values related to new ore targets. However, in the early stages of mineral exploration, DL-based methods face a challenge related to class and sampling imbalance due to scarce mineral deposits, resulting in a lack of enough samples to train, limiting the model’s predictive ability. This work proposed a detailed and systematic framework to address imbalanced data issues with prior geological transfer learning and a weighted loss function. We exploited the abundant pixel information of input variables to develop a pretext geological classification and a feature data extraction task as an initializer for the trainable variables of the neural network. The proposed workflow was tested in a porphyry-rich Yukon (Canada) region and overperformed other state-of-the-art classification algorithms such as random forest, support vector machines, and logistic regression. Moreover, our results were contrasted against different geological reports, where our mineral prospectivity map was coherent with regional and local potential assessments of porphyry-type mineral occurrences. The quantitative metrics with a validation dataset suggested that the proposed method can effectively predict mineral prospective areas in different imbalanced data scenarios.

Mapping the Clinical Pathway for Patients Undergoing Vestibular Schwannoma Resection

The introduction of the electronic health record (EHR) has improved the collection and storage of patient information, enhancing clinical communication and academic research. However, EHRs are limited by data quality and the time-consuming task of manual data extraction. This study aimed to use process mapping to help identify critical data entry points within the clinical pathway for patients with vestibular schwannoma (VS) ideal for structured data entry and automated data collection to improve patient care and research. A 2-stage methodology was used at a neurosurgical unit. Process maps were developed using semi-structured interviews with stakeholders in the management of VS resection. Process maps were then retrospectively validated against EHRs for patients admitted between August 2019 and December 2021, establishing critical data entry points. In the process map development, 20 stakeholders were interviewed. Process maps were validated against EHRs of 36 patients admitted for VS resection. Operative notes, surgical inpatient reviews (including ward rounds), and discharge summaries were available for all patients, representing critical data entry points. Areas for documentation improvement were in the preoperative clinics (30/36; 83.3%), preoperative skull base multidisciplinary team (32/36; 88.9%), postoperative follow-up clinics (32/36; 88.9%), and postoperative skull base multidisciplinary team meeting (29/36; 80.6%). This is a first use to our knowledge of a 2-stage methodology for process mapping the clinical pathway for patients undergoing VS resection. We identified critical data entry points that can be targeted for structured data entry and for automated data collection tools, positively impacting patient care and research.

Case study identification with GPT-4 and implications for mapping studies

Context:Rainer and Wohlin showed that case studies are not well understood by reviewers and authors and thus they say that a given research is a case study when it is not. Objective:Rainer and Wohlin proposed a smell indicator (inspired by code smells) to identify case studies based on the frequency of occurrences of words, which performed better than human classifiers. With the emergence of ChatGPT, we evaluate ChatGPT to assess its performance in accurately identifying case studies. We also reflect on the results’ implications for mapping studies, specifically data extraction. Method:We used ChatGPT with the model GPT-4 to identify case studies and compared the result with the smell indicator for precision, recall, and accuracy. Results:GPT-4 and the smell indicator perform similarly, with GPT-4 performing slightly better in some instances and the smell indicator (SI) in others. The advantage of GPT-4 is that it is based on the definition of case studies and provides traceability on how it reaches its conclusions. Conclusion:As GPT-4 performed well on the task and provides traceability, we should use and, with that, evaluate it on data extraction tasks, supporting us as authors.

Data extraction error in pharmaceutical versus non-pharmaceutical interventions for evidence synthesis: Study protocol for a crossover trial

BackgroundData extraction is the foundation for research synthesis evidence, while data extraction errors frequently occur in the literature. An interesting phenomenon was observed that data extraction error tend to be more common in trials of pharmaceutical interventions compared to non-pharmaceutical ones. The elucidation of which would have implications for guidelines, practice, and policy. Methods and analysesWe propose a crossover, multicenter, investigator-blinded trial to elucidate the potential variants on the data extraction error rates. Eligible 90 participants would be 2nd year or above post-graduate students (e.g., masters, doctoral program). Participants will be randomized to one of the two groups to complete pre-defined data extraction tasks: 1) group A will contain 10 randomized controlled trials (RCTs) of pharmaceutical interventions; 2) group B will contain 10 RCTs of non-pharmaceutical interventions. Participants who finish the data extraction would then be assigned to the alternative group for another round of data extraction after a 30 min washout period. Finally, those participants assigned to A or B group will be further 1:1 randomly matched based on a random-sequenced number for the double-checking process on the extracted data. The primary outcome will be the data extract error rates of the pharmaceutical intervention group and non-pharmaceutical group, before the double-checking process, in terms of the cell level, study level, and participant level. The secondary outcome will be the data error rates of the pharmaceutical intervention group and non-pharmaceutical group after the double-checking process, again, in terms of the cell level, study level, and participant level. A generalized linear mixed effects model (based on the above three levels) will be used to estimate the potential differences in the error rates, with a log link function for binomial data. Subgroup analyses will account for the experience of individuals on systematic reviews and the time used for the data extraction. DiscussionThis trial will provide useful evidence for further systematic review of data extraction practices, improved data extraction strategies, and better guidelines. Trial registrationChinese Clinical Trial Register Center (Identifier: ChiCTR2200062206).

The Future of INCOG (Is Now).

The Future of INCOG (Is Now).

CSF α-Synuclein and Tau as Biomarkers for Dementia With Lewy Bodies: A Systematic Review and Meta-analysis.

This study investigated whether α-synuclein and tau in cerebrospinal fluid (CSF) can be used as biomarkers to diagnose dementia with Lewy bodies (DLB). We retrieved 3303 studies with "Dementia with Lewy bodies," "α-synuclein," and "tau" as keywords. We formulated screening criteria, and 2 researchers completed the screening, quality evaluation, and data extraction tasks. Finally, 35 studies related to tau, and 14 studies related to α-synuclein were included. Review Manager 5.4 and Stata16 were used for meta-analysis. Subgroup, sensitivity, and meta-regression analyses were performed to identify sources of heterogeneity and strengthen the results. Compared with the control group, DLB patients showed significantly higher CSF levels of tau [weighted mean difference=81.36 (59.82, 102.91); Z =7.40; P <0.00001], and lower CSF levels of α-synuclein [weighted mean difference=-95.25 (-162.02, -28.48); Z =2.80; P =0.005]. Mini-Mental State Examination (MMSE) score, male ratio, and disease duration were not sources of heterogeneity on subgroup and meta-regression analyses. Sensitivity analysis revealed no significant differences. Higher levels of tau and lower levels of α-synuclein were found in the CSF of patients with DLB compared with the control group. Therefore, CSF tau and α-synuclein levels may be diagnostic biomarkers for DLB.

A Systematic Review of Personalized Health Applications through Human-Computer Interactions (HCI) on Cardiovascular Health Optimization.

Introduction: Currently, the deployment of human–computer interactive technologies to provide personalized care has grown and immensely taken shape in most healthcare settings. With the increasing growth of the internet and technology, personalized health interventions including smartphones, associated apps, and other interventions demonstrate prowess in various health fields, including cardiovascular management. This systematic review thus examines the effectiveness of various human–computer interactions technologies through telehealth (mainly eHealth) towards optimizing the outcomes in cardiovascular treatment. Methods: A comprehensive search of MEDLINE, EMBASE, and CINAHL databases using key terms was conducted from 2000 to November 2021 to identify suitable studies that explored the use of human–computer interaction technologies to provide a personalized care approach to facilitate bolstered outcomes for cardiovascular patients, including the elderly. The included studies were assessed for quality and risk of bias, and the authors undertook a data extraction task. Results: Ten studies describing the use of a mix of personalized health app (mHealth) interventions were identified and included in the study. Among the included studies, nine of them were randomized trials. All of the studies demonstrated the effectiveness of various personalized health interventions in maximizing the benefits of cardiovascular disease treatment. Conclusions: Personalized health application interventions through precision medicine has great potential to boost cardiovascular disease management outcomes, including rehabilitation. Fundamentally, since each intervention’s focus might differ based on the disease and outcome preference, it is recommended that more research be done to tailor the interventions to specific disease and patient outcome expectations.

Data switching method among heterogeneous power information system databases based on knowledge graph

AbstractIn order to solve the problems of long data switching time, low information utilization rate, and low switching accuracy in traditional data switching methods, a data switching method between heterogeneous power information system databases based on knowledge graphs is proposed. Based on the knowledge graph technology and its ternary structure, the data of the heterogeneous power information system is processed, and the N‐Gram model is used to realize the knowledge extraction, and the data error correction is established through the model. Through ODBC technology to complete the link between heterogeneous power information system databases, and then select the data access interface and determine the data extraction task, and finally realize the data exchange between the heterogeneous power information system databases. Experimental results show that the average communication channel bandwidth during the data switching process is 1.8 Gb/s, the highest value of data switching accuracy reaches 82%, and the information utilization rate after data switching is relatively high, indicating that the method has certain effectiveness.

Natural language query handling using extended knowledge provider system

Extraction of knowledge data from knowledge database using natural language query is a difficult task. Different types of natural language processing (NLP) techniques have been developed to handle this knowledge data extraction task. This paper proposes an automated query-response model termed Extended Automated Knowledge Provider System (EAKPS) that can manage various types of natural language queries from user. The EAKPS uses combination based technique and it can handle assertive, interrogative, imperative, compound and complex type query sentences. The algorithm of EAKPS generates structure query language (SQL) for each natural language query to extract knowledge data from the knowledge database resident within the EAKPS. Extraction of noun or noun phrases is another issue in natural language query processing. Most of the times, determiner, preposition and conjunction are prefixed to a noun or noun phrase and it is difficult to identify the noun/noun phrase with prefix during query processing. The proposed system is able to identify these prefixes and extract exact noun or noun phrases from natural language queries without any manual intervention.

Data extraction methods for systematic review (semi)automation: A living review protocol.

Background: Researchers in evidence-based medicine cannot keep up with the amounts of both old and newly published primary research articles. Conducting and updating of systematic reviews is time-consuming. In practice, data extraction is one of the most complex tasks in this process. Exponential improvements in computational processing speed and data storage are fostering the development of data extraction models and algorithms. This, in combination with quicker pathways to publication, led to a large landscape of tools and methods for data extraction tasks. Objective: To review published methods and tools for data extraction to (semi)automate the systematic reviewing process. Methods: We propose to conduct a living review. With this methodology we aim to do monthly search updates, as well as bi-annual review updates if new evidence permits it. In a cross-sectional analysis we will extract methodological characteristics and assess the quality of reporting in our included papers. Conclusions: We aim to increase transparency in the reporting and assessment of machine learning technologies to the benefit of data scientists, systematic reviewers and funders of health research. This living review will help to reduce duplicate efforts by data scientists who develop data extraction methods. It will also serve to inform systematic reviewers about possibilities to support their data extraction.

Text Mining of the Electronic Health Record: An Information Extraction Approach for Automated Identification and Subphenotyping of HFpEF Patients for Clinical Trials.

Precision medicine requires clinical trials that are able to efficiently enroll subtypes of patients in whom targeted therapies can be tested. To reduce the large amount of time spent screening, identifying, and recruiting patients with specific subtypes of heterogeneous clinical syndromes (such as heart failure with preserved ejection fraction [HFpEF]), we need prescreening systems that are able to automate data extraction and decision-making tasks. However, a major obstacle is the vast amount of unstructured free-form text in medical records. Here we describe an information extraction-based approach that automatically converts unstructured text into structured data, which is cross-referenced against eligibility criteria using a rule-based system to determine which patients qualify for a major HFpEF clinical trial (PARAGON). We show that we can achieve a sensitivity and positive predictive value of 0.95 and 0.86, respectively. Our open-source algorithm could be used to efficiently identify and subphenotype patients with HFpEF and other disorders.

Automated Data Extraction Using Predictive Program Synthesis

In recent years there has been rising interest in the use of programming-by-example techniques to assist users in data manipulation tasks. Such techniques rely on an explicit input-output examples specification from the user to automatically synthesize programs. However, in a wide range of data extraction tasks it is easy for a human observer to predict the desired extraction by just observing the input data itself. Such predictive intelligence has not yet been explored in program synthesis research, and is what we address in this work. We describe a predictive program synthesis algorithm that infers programs in a general form of extraction DSLs (domain specific languages) given input-only examples. We describe concrete instantiations of such DSLs and the synthesis algorithm in the two practical application domains of text extraction and web extraction, and present an evaluation of our technique on a range of extraction tasks encountered in practice.

An Approach for Statistical Data Extraction from Photo Images of Pathological Biopsy Objects

The task of statistical data extraction from photo images of biomedical objects is important in biomedical diagnostics. For example, the analysis of photographic images of aortic valve taken after surgical operation can be used for further medical research. In this case, it is important to define the percentile correlation between the pathological and the macroscopically unchanged tissue. In this work, authors implement different methods for extracting the statistical data from images. The experimental results show the efficiency of the selected methods.

A machine reading system for assembling synthetic paleontological databases.

Many aspects of macroevolutionary theory and our understanding of biotic responses to global environmental change derive from literature-based compilations of paleontological data. Existing manually assembled databases are, however, incomplete and difficult to assess and enhance with new data types. Here, we develop and validate the quality of a machine reading system, PaleoDeepDive, that automatically locates and extracts data from heterogeneous text, tables, and figures in publications. PaleoDeepDive performs comparably to humans in several complex data extraction and inference tasks and generates congruent synthetic results that describe the geological history of taxonomic diversity and genus-level rates of origination and extinction. Unlike traditional databases, PaleoDeepDive produces a probabilistic database that systematically improves as information is added. We show that the system can readily accommodate sophisticated data types, such as morphological data in biological illustrations and associated textual descriptions. Our machine reading approach to scientific data integration and synthesis brings within reach many questions that are currently underdetermined and does so in ways that may stimulate entirely new modes of inquiry.

Structured Information Extraction from Medical Texts in Bulgarian

Abstract This paper presents an approach for Information Extraction (IE) from Patient Records (PRs) in Bulgarian. The specific terminology and lack of resources in electronic format are some of the obstacles that make the task of current patient status data extraction in a structured format quite challenging. The usage of N-grams, collocations and words’ distances allows us to cope with this problem and to extract automatically the attribute-value pairs with relatively high precision.

Directly e-mailing authors of newly published papers encourages community curation

Much of the data within Model Organism Databases (MODs) comes from manual curation of the primary research literature. Given limited funding and an increasing density of published material, a significant challenge facing all MODs is how to efficiently and effectively prioritize the most relevant research papers for detailed curation. Here, we report recent improvements to the triaging process used by FlyBase. We describe an automated method to directly e-mail corresponding authors of new papers, requesting that they list the genes studied and indicate (‘flag’) the types of data described in the paper using an online tool. Based on the author-assigned flags, papers are then prioritized for detailed curation and channelled to appropriate curator teams for full data extraction. The overall response rate has been 44% and the flagging of data types by authors is sufficiently accurate for effective prioritization of papers. In summary, we have established a sustainable community curation program, with the result that FlyBase curators now spend less time triaging and can devote more effort to the specialized task of detailed data extraction.Database URL: http://flybase.org/

Observing effectiveness of pathology ordering controls in emergency departments.

To demonstrate application of data integration technology for observing the effectiveness of interventions to control pathology orders in Emergency Departments. Doctors frequently need to order blood tests in the Emergency Departments as a part of diagnostic set up in Emergency Departments. However, pathology test ordering is excessive and often unnecessary. The excessive ordering of blood test places a significant financial burden on our health care system. It also causes undue discomfort and worry to the patients. There are many interventions employed to control pathology ordering in Emergency Departments. The analysis of effectiveness of interventions is required for improving clinical practices in Emergency Departments. However, the collection and extraction of data on the effects of intervention can be very costly and time consuming. Therefore, there is a need of a technology-based solution to access, query and analyse data residing across different sources. The research aims to determine efficacy of an intervention called the "Traffic Light System" through a pathology request form used to control the pathology ordering in one adult hospital emergency department. Health Data Integration (HDI) technology was implemented to link and query the data residing at different source systems i.e. pathology and ED information system. The data was extracted from the Emergency Department Information System at an adult tertiary hospital in Queensland. Twenty weeks of pre-intervention data was collected. Twenty weeks of post-intervention data was collected after 32-week transition interval. The data for pre-intervention, transition and post-intervention period was analysed to assess the effectiveness of the intervention in reducing commonly ordered pathology tests such as Full Blood Counts (FBC) and Erythrocyte Sedimentation Rate (ESR). The total number of FBC tests ordered in the pre-intervention period fell slightly in the post-intervention period (mean 42.3 vs 38.1 per 100 patients). The total number Erythrocyte Sedimentation Rate tests showed a significant declining trend as a result of ED intervention (2.5 vs 1.4 per 100 patients, p=0.001). HDI completed the task of data extraction, manipulation and querying in seconds. A manual check of a sample of 200 pathology test orders shows 95.5% sensitivity, which is considered accurate enough for this purpose. Pathology ordering can be reduced using sustainable protocols. This work has demonstrated HDI capability to extract and link pathology data efficiently to evaluate an ED intervention.

Rule-based information extraction from patients’ clinical data

The paper describes a rule-based information extraction (IE) system developed for Polish medical texts. We present two applications designed to select data from medical documentation in Polish: mammography reports and hospital records of diabetic patients. First, we have designed a special ontology that subsequently had its concepts translated into two separate models, represented as typed feature structure (TFS) hierarchies, complying with the format required by the IE platform we adopted. Then, we used dedicated IE grammars to process documents and fill in templates provided by the models. In particular, in the grammars, we addressed such linguistic issues as: ambiguous keywords, negation, coordination or anaphoric expressions. Resolving some of these problems has been deferred to a post-processing phase where the extracted information is further grouped and structured into more complex templates. To this end, we defined special heuristic algorithms on the basis of sample data. The evaluation of the implemented procedures shows their usability for clinical data extraction tasks. For most of the evaluated templates, precision and recall well above 80% were obtained.

Automatically maintaining navigation sequences for querying semi-structured web sources

A substantial subset of Web data has an underlying structure. For instance, the pages obtained in response to a query executed through a Web search form are usually generated by a program that accesses structured data in a local database, and embeds them into an HTML template. For software programs to gain full benefit from these “semi-structured” Web sources, wrapper programs must be built to provide a “machine-readable” view over them. Since Web sources are autonomous, they may experience changes that invalidate the current wrapper, thus automatic maintenance is an important issue. Wrappers must perform two tasks: navigating through Web sites and extracting structured data from HTML pages. While several works have addressed the automatic maintenance of data extraction tasks, the problem of maintaining the navigation sequences remains unaddressed to the best of our knowledge. In this paper, we propose a set of novel techniques to fill this gap.

Structured Data Extraction from the Web Based on Partial Tree Alignment

This paper studies the problem of structured data extraction from arbitrary Web pages. The objective of the proposed research is to automatically segment data records in a page, extract data items/fields from these records, and store the extracted data in a database. Existing methods addressing the problem can be classified into three categories. Methods in the first category provide some languages to facilitate the construction of data extraction systems. Methods in the second category use machine learning techniques to learn wrappers (which are data extraction programs) from human labeled examples. Manual labeling is time-consuming and is hard to scale to a large number of sites on the Web. Methods in the third category are based on the idea of automatic pattern discovery. However, multiple pages that conform to a common schema are usually needed as the input. In this paper, we propose a novel and effective technique (called DEPTA) to perform the task of Web data extraction automatically. The method consists of two steps: 1) identifying individual records in a page and 2) aligning and extracting data items from the identified records. For step 1, a method based on visual information and tree matching is used to segment data records. For step 2, a novel partial alignment technique is proposed. This method aligns only those data items in a pair of records that can be aligned with certainty, making no commitment on the rest of the items. Experimental results obtained using a large number of Web pages from diverse domains show that the proposed two-step technique is highly effective.