A comparison of virtual reality and three-dimensional multiplanar educational methods for student learning of cone beam computed tomography interpretations.

OBJECTIVES: The purpose of this study was to compare student learning of cone beam computed tomography (CBCT) interpretation using immersive virtual reality (VR) and three-dimensional multiplanar (MP) reconstructions. METHODS: Sixty first-year dental students were randomly allocated to two groups, VR and MP, and underwent a one-on-one educational intervention to identify anatomic structures using CBCT data. All participants completed three multiple-choice questionnaires (MCQs) before (T1), immediately after (T2), and 2 weeks following (T3) the intervention. Additionally, pre-survey, post-survey, NASA Task Load Index (NASA-TLX), and presence questionnaires were completed. Analysis of objective measures of performance on MCQs and subjective data from the questionnaires was completed (α = 0.05). RESULTS: There was a significant increase in test performance and informational recall between T1-T2 and T1-T3 for VR and MP groups (p < 0.001). However, there were no significant differences in performance on MCQs between T2 and T3. Analysis of the Presence questionnaire indicated that the VR group felt decreased distraction (p = 0.013), increased realism (p = 0.035), and increased involvement (p = 0.047) during the educational intervention when compared with the MP group. Analysis of the NASA-TLX indicated that the VR group experienced more physical demand (p < 0.01) but similar cognitive demand when compared with the MP group. Qualitative responses indicated that the VR group had a more dynamic sense of visualization and manipulation compared to the MP group. CONCLUSION: Results from this study show that VR is as effective as traditional MP methods of CBCT interpretation learning. Further benefits of VR educational intervention include increased involvement, realism and less distraction.

Synthesis of Nano Sulfur/Chitosan-Copper Complex and Its Nematicidal Effect against Meloidogyne incognita In Vitro and on Coffee Pots.

Sulfur is one of the inorganic elements used by plants to develop and produce phytoalexin to resist certain diseases. This study reported a method for preparing a material for plant disease resistance. Sulfur nanoparticles (SNPs) stabilized in the chitosan-Cu2+ (CS-Cu2+) complex were synthesized by hydrolysis of Na2S2O3 in an acidic medium. The obtained SNPs/CS-Cu2+ complex consisting of 0.32% S, 4% CS, and 0.7% Cu (w/v), contained SNPs with an average size of ~28 nm as measured by transmission electron microscopy images. The X-ray diffraction pattern of the SNPs/CSCu2+ complex showed that SNPs had orthorhombic crystal structures. Interaction between SNPs and the CS-Cu2+ complex was also investigated by ultraviolet-visible. Results in vitro nematicidal effect of materials against Meloidogyne incognita showed that SNPs/CS-Cu2+ complex was more effective in killing second-stage juveniles (J2) nematodes and inhibiting egg hatching than that of CS and CS-Cu2+ complex. The values of LC50 in killing J2 nematodes and EC50 in inhibiting egg hatching of SNPs/CS-Cu2+ complex were 75 and 51 mg/l, respectively. These values were lower than those of CS and the CS-Cu2+ complex. The test results on the nematicidal effect against M. incognita on coffee pots showed that the SNPs/CS-Cu2+ complex was 100% effective at a concentration of 150 mg/l. Therefore, the SNPs/CS-Cu2+ complex could be considered as a biochemical material with potential for agricultural applications to control root-knot nematodes.

Preparation of water-soluble chitosan oligosaccharides by oxidative hydrolysis of chitosan powder with hydrogen peroxide.

Chitosan (CS) is only soluble in weak acid medium, thereby limiting its wide utilisation in the field of biomedicine, food, and agriculture. In this report, we present a method for preparing water-soluble CS oligosaccharides (COSs) at high concentration (∼10%, w/v) via the oxidative hydrolysis of CS powder with molecular weight (Mw) ∼90,000 g/mol) in 2% H2O2 solution at ambient temperature by a two-step process, namely, the heterogeneous hydrolysis step and homogeneous hydrolysis step. The resultant COSs were characterised by gel permeation chromatography (GPC), fourier transforms infrared spectroscopy (FT-IR), ultraviolet-visible spectroscopy (UV-Vis), proton nuclear magnetic resonance spectroscopy (1H NMR) and X-ray diffraction (XRD) spectroscopy. The resulting products were composed of COSs (Mw of 2000-6600 g/mol) that were completely soluble in water. The results also indicated that the structure of COSs was almost unchanged compared with the original CS unless Mw was low. Accordingly, COSs with low Mw (∼2000 g/mol) and high concentration (10%, w/v) could be effectively prepared by the oxidative hydrolysis of CS powder using hydrogen peroxide under ambient conditions.

Isolated torsion of fallopian tube with associated torsed paratubal cystadenofibroma: A case report.

Cystadenofibromas (CAF) and adenofibromas (AF) are rare benign gynecologic neoplasms of epithelial origin. They can be composed predominantly of solid fibrous tissue, adenofibromas, or contain cystic components, as seen in cystadenofibromas; which can be further classified by their epithelial cell type (serous, mucinous, etc.). They most commonly arise from the ovary; however, CAF/AF associated with the fallopian tube have also been reported. CAF/AF is typically asymptomatic and found incidentally. Like other adnexal and ovarian masses, patients are at increased risk of ovarian/adnexal torsion, with increasing risk with lesion size. Herein, we present the case of a patient with isolated torsion of a fallopian tube with associated torsed paratubal serous cystadenofibroma. To our knowledge, this is the first reported case of isolated torsion of a fallopian tube with associated para-tubal cystadenofibroma.

Enlarging premaxillary mass: A unique manifestation of extranodal diffuse B-cell lymphoma.

Lymphomas are the most common nonepithelial malignancy in the head and neck region. Among these, non-Hodgkin Lymphoma (NHL) is the most prevalent, and diffuse large B-cell lymphoma (DLBCL) is the most common histologic subtype. NHL is known for its propensity for extranodal involvement, which can affect any anatomical location. The presence of perineural spread is frequently encountered in head and neck malignancies, including lymphomas. We report a case of a 40-year-old male with an enlarging infraorbital facial mass with associated erythema, pain, and paresthesia, which was subsequently found to be extranodal DLBCL with retrograde perineural spread along the infraorbital nerve.

The adhesion GPCRs CELSR1-3 and LPHN3 engage G proteins via distinct activation mechanisms.

Adhesion G protein-coupled receptors (aGPCRs) are a large GPCR class that direct diverse fundamental biological processes. One prominent mechanism for aGPCR agonism involves autoproteolytic cleavage, which generates an activating, membrane-proximal tethered agonist (TA). How universal this mechanism is for all aGPCRs is unclear. Here, we investigate G protein induction principles of aGPCRs using mammalian latrophilin 3 (LPHN3) and cadherin EGF LAG-repeat 7-transmembrane receptors 1-3 (CELSR1-3), members of two aGPCR families conserved from invertebrates to vertebrates. LPHNs and CELSRs mediate fundamental aspects of brain development, yet CELSR signaling mechanisms are unknown. We find that CELSR1 and CELSR3 are cleavage deficient, while CELSR2 is efficiently cleaved. Despite differential autoproteolysis, CELSR1-3 all engage GαS, and CELSR1 or CELSR3 TA point mutants retain GαS coupling activity. CELSR2 autoproteolysis enhances GαS coupling, yet acute TA exposure alone is insufficient. These studies support that aGPCRs signal via multiple paradigms and provide insights into CELSR biological function.

The adhesion GPCRs CELSR1-3 and LPHN3 engage G proteins via distinct activation mechanisms.

Adhesion GPCRs (aGPCRs) are a large GPCR class that direct diverse fundamental biological processes. One prominent mechanism for aGPCR agonism involves autoproteolytic cleavage, which generates an activating, membrane-proximal tethered agonist (TA). How universal this mechanism is for all aGPCRs is unclear. Here, we investigate G protein induction principles of aGPCRs using mammalian LPHN3 and CELSR1-3, members of two aGPCR families conserved from invertebrates to vertebrates. LPHNs and CELSRs mediate fundamental aspects of brain development, yet CELSR signaling mechanisms are unknown. We found that CELSR1 and CELSR3 are cleavage-deficient, while CELSR2 is efficiently cleaved. Despite differential autoproteolysis, CELSR1-3 all engage GαS, and CELSR1 or CELSR3 TA point mutants retain GαS coupling activity. CELSR2 autoproteolysis enhances GαS coupling, yet acute TA exposure alone is insufficient. These studies support that aGPCRs signal via multiple paradigms and provide insights into CELSR biological function.

Student learning of head and neck anatomy using cone beam computed tomography and immersive virtual reality.

OBJECTIVES: The purpose of this study was to evaluate the efficacy of student learning of anatomy and 3D imaging concepts using cone beam computed tomography (CBCT) and immersive virtual reality (VR) technology. METHODOLOGY: Ninety (n = 90) first year dental students with no previous experience in 3D imaging were recruited. All participants completed a 10-item, multiple-choice questionnaire (MCQ) and a pre-survey prior to the educational intervention. Following a brief video orientation on CBCT and anatomy, each participant underwent a one-on-one educational intervention using immersive VR with calibrated instructors to identify head and neck anatomic structures using a VR/CBCT educational tool. Immediately following the intervention, all participants completed a postsurvey, a second MCQ, NASA task load index and presence questionnaires. Participants completed a third MCQ 2 weeks following the intervention. Analysis of objective measures of performance on MCQ's (p < 0.05) and subjective data from the questionnaires was completed. RESULTS: The students doubled their mean test scores 2.45 ± 1.274 to 5.99 ± 1.576 on MCQ's immediately following the educational intervention (p < 0.05). The significant increase in the MCQ test scores was maintained after 2 weeks, 5.73 ± 1.721 (p < 0.05). There were no gender differences in student test performance. Students rated the immersive VR/CBCT educational intervention experience highly for control, sensory, and realism factors with minimal distraction and frustration factors. CONCLUSION: Results from this study show that immersive VR/CBCT educational intervention improved test performance and contributed to information recall in students. Further benefits reported by participants include the sense of presence and increased engagement using immersive VR.

Physico-chemical properties and transboundary transport of PM2.5 in Bien Hoa City, Dong Nai Province, Southeastern Vietnam.

Air pollution is one of the most concerning environmental issues, wherein PM2.5 concentration plays an important role. This study monitored and evaluated the PM2.5 concentration trends in Bien Hoa City, Dong Nai Province, Vietnam. Twenty 24-h PM2.5 samples were continuously collected during the rainy (15 Oct. to 25 Oct. 2021) and dry (19 Mar. to 29 Mar. 2022) seasons. The PM2.5 samples were analyzed by scanning electron microscopy to determine the surface pattern and size distributions were analyzed using ImageJ software. The water-soluble fractions of 15 trace metal(oid)s concentrations (Al, Cu, Ni, K, Ca, Co, Mn, Cr, As, Zn, Pb, Cd, Na, Fe, and Mg) bound to PM2.5 were analyzed by inductively coupled plasma-mass spectrometry. The results showed that the 24-h fine fraction PM2.5 concentrations were 24.1 ± 12.2 µg/m3 and 63.0 ± 18.7 µg/m3 in the rainy and dry seasons, respectively. The results indicate that the size distributions of the particles of 2.0-2.5 µm are minor, and the majority are ultrafine particles with aerodynamic diameter ≤ 1.0 µm. Overall, the mass concentration level of the water-soluble fraction of trace metal(oid)s in PM2.5 in the rainy season was higher than that in the dry season. Among these, Ca, K, and Na were the most abundant earth crustal elements in PM2.5 in the rainy and dry seasons, accounting for 85% and 41.2% of the total trace element concentrations, respectively. The major sources of PM2.5 are local and regional sources of thermal power plants, industrial parks, and waterborne transportation (domestic rivers and marine). The activities undertaken to remove Agent Orange (e.g., soil excavation, transportation, and rotary kiln incinerators) at the Bien Hoa airbase area also cause increases in the PM2.5 level in the atmosphere of Bien Hoa City.

A case report of a multi-arterial dural arteriovenous fistula.

Dural arteriovenous fistulas are rare intracranial vascular malformations with a propensity for hemorrhage. The Cognard classification system is the most widespread classification system wherein type IIB through V must be promptly treated to avoid the risk of hemorrhage. The case presented herein reports a 71-year-old male presenting with vague non-hemorrhagic neurologic deficits found to have a Cognard type III dural arteriovenous fistula with multiple arterial feeders. Although quite obvious in retrospect, a DAVF can be missed even by an astute radiologist. This should be considered a "never miss" diagnosis as it carries a risk of intracranial hemorrhage and death.

Ovarian Torsion: Presentation and Management in a Pediatric Patient.

Background: Adnexal torsion is the fifth most common gynecologic emergency accounting for approximately 20 to 30% of ovarian surgeries in pediatric patients. Case: The patient is a ten-year-old female who presented to the emergency room for severe left lower quadrant abdominal pain. On presentation, she was hemodynamically stable with an acute abdomen. A transabdominal ultrasound showed a predominantly anechoic structure measuring up to 5.6 cm without definitive Doppler flow, concerning for a large cyst causing ovarian torsion. Gynecology was consulted, and the patient underwent a diagnostic laparoscopy, aspiration of the left ovarian cyst, and left ovarian detorsion. Pathology results were consistent with benign cystic contents. Conclusion: Appropriate diagnosis and timely surgical gynecological intervention allowed this pediatric patient to salvage and preserve ovarian function.

Transmission channels between financial development and CO 2 emissions: A global perspective.

This paper aims to investigate the direct and indirect effects of financial development on CO2 emissions, using a global sample of 100 countries from 1990 - 2012. Our main contribution to the literature lies in the identification and explanation of possible transmission channels that allow financial development to affect environmental quality. The paper employs 2SLS and 3SLS estimators to investigate these channels. Empirical results confirm the positive direct effect of financial development on environmental degradation. Development of the financial system also gives rise to more energy demand and consequently leads to more pollutant emissions. Besides, there is evidence about a trade-off between income inequality and environmental quality. Financial development can help redistribute income more effectively. However, high living standards will put pressure on environmental conservation. The paper also considers the nonlinear effects of financial development on carbon emission rates. Only a small proportion of the population receive the benefits at the early stages of financial development. After a certain amount of time, financial development benefits a more significant part of the population and reduces income inequality.

Development of a Multistep Hypertension Quality Improvement Program in an Academic General Medicine Practice.

Hypertension is a common problem and a major risk factor for cardiovascular disease. It is unclear whether efforts to control blood pressure (BP) known to be effective in integrated healthcare systems can be successfully implemented in an academic setting. We describe our experience implementing a multistep quality improvement program within an academic general medicine practice aimed at improving BP among patients with uncontrolled hypertension. Ensuring medical assistants were correctly measuring BP provided the basis for accurate data entry into the electronic medical record (EMR); our EMR-based registry data allowed us to feedback primary care provider (PCP) level data on BP control for panel management, which resulted in improvements in BP for a substantial proportion of patients, particularly for those with more practice visits. However, due to PCP, patient, and system barriers, our initial attempt to integrate a pharmacist into our team for hypertension management was only successful for a small number of patients who engaged in pharmacist in-person visits. Future improvement efforts will focus on addressing the barriers to more intensive BP management, integrating lessons from this experience. As chronic disease management shifts to a population-based model, team change will be a necessary component for achieving clinical improvement.

Estimation of Radionuclide Concentrations and Average Annual Committed Effective Dose due to Ingestion for the Population in the Red River Delta, Vietnam.

Radioactivity concentrations of nuclides of the 232Th and 238U radioactive chains and 40K, 90Sr, 137Cs, and 239+240Pu were surveyed for raw and cooked food of the population in the Red River delta region, Vietnam, using α-, γ-spectrometry, and liquid scintillation counting techniques. The concentration of 40K in the cooked food was the highest compared to those of other radionuclides ranging from (23 ± 5) (rice) to (347 ± 50) Bq kg-1 dw (tofu). The 210Po concentration in the cooked food ranged from its limit of detection (LOD) of 5 mBq kg-1 dw (rice) to (4.0 ± 1.6) Bq kg-1 dw (marine bivalves). The concentrations of other nuclides of the 232Th and 238U chains in the food were low, ranging from LOD of 0.02 Bq kg-1 dw to (1.1 ± 0.3) Bq kg-1 dw. The activity concentrations of 90Sr, 137Cs, and 239+240Pu in the food were minor compared to that of the natural radionuclides. The average annual committed effective dose to adults in the study region was estimated and it ranged from 0.24 to 0.42 mSv a-1 with an average of 0.32 mSv a-1, out of which rice, leafy vegetable, and tofu contributed up to 16.2%, 24.4%, and 21.3%, respectively. The committed effective doses to adults due to ingestion of regular diet in the Red River delta region, Vietnam are within the range determined in other countries worldwide. This finding suggests that Vietnamese food is safe for human consumption with respect to radiation exposure.

Is Multiclass Automatic Text De-Identification Worth the Effort?

OBJECTIVES: Automatic de-identification to remove protected health information (PHI) from clinical text can use a "binary" model that replaces redacted text with a generic tag (e.g., ""), or can use a "multiclass" model that retains more class information (e.g., ""). Binary models are easier to develop, but result in text that is potentially less informative. We investigated whether building a multiclass de-identification is worth the extra effort. METHODS: Using the 2014 i2b2 dataset, we compared the accuracy and impact on document readability of two models. In the first experiment, we generated one binary and two multiclass versions trained with the same machine-learning algorithm Conditional Random Field (CRF). Accuracy (recall, precision, f-score) and secondary metrics (e.g, training time, testing time, minimum memory required) were measured. In the second experiment, three reviewers accessed the readability of two redacted documents using the binary and multiclass methods. We estimated a pooled Kappa to estimate the inter-rater agreement. RESULTS: The multiclass model did not demonstrate a clear accuracy advantage, with lower recall (-1.9%) and only slightly better precision (+0.6%), despite requiring additional computing resources. Three raters reached a very high agreement (Kappa = 0.975, 95% Confidence Interval (0.946, 1.00), p < 0.0001) that both binary and multiclass models have the same impact on document readability. CONCLUSIONS: This study suggests that the development of more sophisticated classification of PHI may not be worth the effort in terms of both system accuracy and the usefulness of the output.

Molecular absorption measurements with an optical fibre coupled array of ultra-violet light-emitting diodes.

A photometric detector based on eight different light-emitting diodes covering the ultraviolet range from 255 nm to 350 nm is described. These are coupled with fused silica optical fibres to a conventional cuvette with 1 cm optical path length or to a low volume flow through cell for detection in high-performance liquid chromatography. Photodiodes are employed for the measurement of the transmitted intensity as well as of a reference signal and the photocurrents are processed with a log-ratio amplifier to obtain a voltage proportional to absorbance values. The wavelength desired for the measurement at hand is selected by electronically switching on the requisite light-emitting diode. The detector was found to have a low noise level of 80 µAU. In batch-wise measurements as well as in detection for high-performance liquid chromatography dynamic ranges of 2-3 orders of magnitude were possible. Reproducibilities in peak areas for the latter application were better than 1%.

The UAB Informatics Institute and 2016 CEGS N-GRID de-identification shared task challenge.

Clinical narratives (the text notes found in patients' medical records) are important information sources for secondary use in research. However, in order to protect patient privacy, they must be de-identified prior to use. Manual de-identification is considered to be the gold standard approach but is tedious, expensive, slow, and impractical for use with large-scale clinical data. Automated or semi-automated de-identification using computer algorithms is a potentially promising alternative. The Informatics Institute of the University of Alabama at Birmingham is applying de-identification to clinical data drawn from the UAB hospital's electronic medical records system before releasing them for research. We participated in a shared task challenge by the Centers of Excellence in Genomic Science (CEGS) Neuropsychiatric Genome-Scale and RDoC Individualized Domains (N-GRID) at the de-identification regular track to gain experience developing our own automatic de-identification tool. We focused on the popular and successful methods from previous challenges: rule-based, dictionary-matching, and machine-learning approaches. We also explored new techniques such as disambiguation rules, term ambiguity measurement, and used multi-pass sieve framework at a micro level. For the challenge's primary measure (strict entity), our submissions achieved competitive results (f-measures: 87.3%, 87.1%, and 86.7%). For our preferred measure (binary token HIPAA), our submissions achieved superior results (f-measures: 93.7%, 93.6%, and 93%). With those encouraging results, we gain the confidence to improve and use the tool for the real de-identification task at the UAB Informatics Institute.

Extractive text summarization system to aid data extraction from full text in systematic review development.

OBJECTIVES: Extracting data from publication reports is a standard process in systematic review (SR) development. However, the data extraction process still relies too much on manual effort which is slow, costly, and subject to human error. In this study, we developed a text summarization system aimed at enhancing productivity and reducing errors in the traditional data extraction process. METHODS: We developed a computer system that used machine learning and natural language processing approaches to automatically generate summaries of full-text scientific publications. The summaries at the sentence and fragment levels were evaluated in finding common clinical SR data elements such as sample size, group size, and PICO values. We compared the computer-generated summaries with human written summaries (title and abstract) in terms of the presence of necessary information for the data extraction as presented in the Cochrane review's study characteristics tables. RESULTS: At the sentence level, the computer-generated summaries covered more information than humans do for systematic reviews (recall 91.2% vs. 83.8%, p<0.001). They also had a better density of relevant sentences (precision 59% vs. 39%, p<0.001). At the fragment level, the ensemble approach combining rule-based, concept mapping, and dictionary-based methods performed better than individual methods alone, achieving an 84.7% F-measure. CONCLUSION: Computer-generated summaries are potential alternative information sources for data extraction in systematic review development. Machine learning and natural language processing are promising approaches to the development of such an extractive summarization system.

PDF text classification to leverage information extraction from publication reports.

OBJECTIVES: Data extraction from original study reports is a time-consuming, error-prone process in systematic review development. Information extraction (IE) systems have the potential to assist humans in the extraction task, however majority of IE systems were not designed to work on Portable Document Format (PDF) document, an important and common extraction source for systematic review. In a PDF document, narrative content is often mixed with publication metadata or semi-structured text, which add challenges to the underlining natural language processing algorithm. Our goal is to categorize PDF texts for strategic use by IE systems. METHODS: We used an open-source tool to extract raw texts from a PDF document and developed a text classification algorithm that follows a multi-pass sieve framework to automatically classify PDF text snippets (for brevity, texts) into TITLE, ABSTRACT, BODYTEXT, SEMISTRUCTURE, and METADATA categories. To validate the algorithm, we developed a gold standard of PDF reports that were included in the development of previous systematic reviews by the Cochrane Collaboration. In a two-step procedure, we evaluated (1) classification performance, and compared it with machine learning classifier, and (2) the effects of the algorithm on an IE system that extracts clinical outcome mentions. RESULTS: The multi-pass sieve algorithm achieved an accuracy of 92.6%, which was 9.7% (p<0.001) higher than the best performing machine learning classifier that used a logistic regression algorithm. F-measure improvements were observed in the classification of TITLE (+15.6%), ABSTRACT (+54.2%), BODYTEXT (+3.7%), SEMISTRUCTURE (+34%), and MEDADATA (+14.2%). In addition, use of the algorithm to filter semi-structured texts and publication metadata improved performance of the outcome extraction system (F-measure +4.1%, p=0.002). It also reduced of number of sentences to be processed by 44.9% (p<0.001), which corresponds to a processing time reduction of 50% (p=0.005). CONCLUSIONS: The rule-based multi-pass sieve framework can be used effectively in categorizing texts extracted from PDF documents. Text classification is an important prerequisite step to leverage information extraction from PDF documents.

Automatically finding relevant citations for clinical guideline development.

OBJECTIVE: Literature database search is a crucial step in the development of clinical practice guidelines and systematic reviews. In the age of information technology, the process of literature search is still conducted manually, therefore it is costly, slow and subject to human errors. In this research, we sought to improve the traditional search approach using innovative query expansion and citation ranking approaches. METHODS: We developed a citation retrieval system composed of query expansion and citation ranking methods. The methods are unsupervised and easily integrated over the PubMed search engine. To validate the system, we developed a gold standard consisting of citations that were systematically searched and screened to support the development of cardiovascular clinical practice guidelines. The expansion and ranking methods were evaluated separately and compared with baseline approaches. RESULTS: Compared with the baseline PubMed expansion, the query expansion algorithm improved recall (80.2% vs. 51.5%) with small loss on precision (0.4% vs. 0.6%). The algorithm could find all citations used to support a larger number of guideline recommendations than the baseline approach (64.5% vs. 37.2%, p<0.001). In addition, the citation ranking approach performed better than PubMed's "most recent" ranking (average precision +6.5%, recall@k +21.1%, p<0.001), PubMed's rank by "relevance" (average precision +6.1%, recall@k +14.8%, p<0.001), and the machine learning classifier that identifies scientifically sound studies from MEDLINE citations (average precision +4.9%, recall@k +4.2%, p<0.001). CONCLUSIONS: Our unsupervised query expansion and ranking techniques are more flexible and effective than PubMed's default search engine behavior and the machine learning classifier. Automated citation finding is promising to augment the traditional literature search.