SEARCH · PubMed Health

Results for “Large Language Models”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 685 records · Page 38Linked to original sources

PubMed2005

Structural semantic interconnections: a knowledge-based approach to word sense disambiguation.

Word Sense Disambiguation (WSD) is traditionally considered an Al-hard problem. A break-through in this field would have a significant impact on many relevant Web-based applications, such as Web information retrieval, improved access to Web services, information extraction, etc. Early approaches to WSD, based on knowledge representation techniques, have been replaced in the past few years by more robust machine learning and statistical techniques. The results of recent comparative evaluations of WSD systems, however, show that these methods have inherent limitations. On the other hand, the increasing availability of large-scale, rich lexical knowledge resources seems to provide new challenges to knowledge-based approaches. In this paper, we present a method, called structural semantic interconnections (SSI), which creates structural specifications of the possible senses for each word in a context and selects the best hypothesis according to a grammar G, describing relations between sense specifications. Sense specifications are created from several available lexical resources that we integrated in part manually, in part with the help of automatic procedures. The SSI algorithm has been applied to different semantic disambiguation problems, like automatic ontology population, disambiguation of sentences in generic texts, disambiguation of words in glossary definitions. Evaluation experiments have been performed on specific knowledge domains (e.g., tourism, computer networks, enterprise interoperability), as well as on standard disambiguation test sets.

Algorithms↗

PubMed2006

Normative data for the Animal, Profession and Letter M Naming verbal fluency tests for Dutch speaking participants and the effects of age, education, and sex.

Previous research has indicated that performance on verbal fluency tests (VFTs) is influenced by language and/or culture. Consequently, normative VFT data for English-speaking people cannot be used for people for whom English is not their first language. The aim of the present study was to provide normative data for the Animal Naming, Profession Naming, and Letter M Naming (four-letter words beginning with the letter M) VFTs for Dutch-speaking populations, based on a large sample (N = 1856) of healthy men and women aged 24-81 years of different educational levels. The results showed that age affected the performance of all VFTs profoundly, but the age effect was not uniform: in the Profession and Letter M Naming VFTs, performance was stable in young adulthood but declined strongly after age 50. In contrast, in the Animal Naming VFT, performance appeared to decline linearly, starting early in life. Furthermore, males had higher scores than females on the Profession Naming VFT, and higher educated participants outperformed their lower educated counterparts on all three VFTs. Regression-based normative data were prepared for the 3 VFTs, and the advantages of using a regression-based normative approach instead of a traditional normative approach are discussed.

Adult↗

PubMed2003

Meta-analysis of corticosteroid treatment in acute myocardial infarction.

Acute and chronic inflammation play a central role in the pathophysiology of atherosclerosis. Corticosteroids are the gold standard anti-inflammatory agent and may have a role in treating acute myocardial infarction. However, concern exists regarding the potential for impaired wound healing and wall thinning. The MEDLINE and PreMEDLINE databases were searched for articles from 1966 through May 2002. A total of 186 articles and 16 English-language publications were identified. A meta-analysis of mortality in controlled trials was performed. Sensitivity analyses and 2 tests for publication bias were used to test the robustness of the results. Sixteen studies involving 3,793 patients were reviewed. Most studies were small (<100 patients) and revealed conflicting efficacy using surrogate outcome measures, such as infarct size. No clear association with myocardial rupture was observed. Meta-analysis of 11 controlled trials (2,646 patients) revealed a 26% decrease in mortality with corticosteroids (odds ratio 0.74, 95% confidence interval [CI] 0.59 to 0.94; p = 0.015). Sensitivity analyses limited to large studies and randomized controlled trials revealed odds ratios of 0.76 (95% CI 0.53 to 1.09) and 0.95 (95% CI 0.72 to 1.26), respectively. Two tests revealed no evidence for publication bias. Thus, the review of available clinical studies demonstrated no harm and a possible mortality benefit of corticosteroids in acute myocardial infarction.

Adrenal Cortex Hormones↗

PubMed2005

Theophylline for prevention of contrast-induced nephropathy: a systematic review and meta-analysis.

BACKGROUND: Contrast-induced nephropathy (CIN) is an important cause of declines in kidney function and is related to greater morbidity, health care costs, and mortality. Adenosine has been proposed to contribute to the pathophysiological process of CIN. We performed a systematic review and meta-analysis of theophylline, an adenosine antagonist, for the prevention of CIN. DATA SOURCES: Studies were identified in all languages by search of MEDLINE (1966 through November 2003), EMBASE (1980 through week 44 [November] of 2003), and the Cochrane Controlled Clinical Trials Register (1996 through November 2003) databases and selected conference proceedings. STUDY SELECTION: We searched for randomized controlled trials comparing theophylline vs control in patients receiving radiocontrast media for angiography or computed tomography. DATA EXTRACTION: Our primary outcome measures were the risk of CIN, the difference in serum creatinine levels between theophylline and control groups at 48 hours and need for dialysis. DATA SYNTHESIS: Nine randomized controlled trials involving 585 patients were identified and included for analysis. Theophylline protocols and definitions of CIN varied across studies. There was evidence of heterogeneity of results across trials (Q = 9.77; P = .08); therefore, pooled values require cautious interpretation. The overall pooled odds ratio (OR) using a conservative random-effects model was 0.40 (95% confidence interval [CI], 0.14 to 1.16; P = .09) indicating a trend toward reduction in the incidence of CIN with theophylline use. The pooled estimate for the difference in 48-hour serum creatinine levels between the theophylline and control groups was -0.17 mg/dL (95% CI, -0.28 to -0.06 mg/dL) (-15.2 micromol/L [95% CI, -24.6 to -5.7 micromol/L]) (P = .002), indicating that theophylline may be protective in CIN. The incidence of CIN requiring dialysis was uncommon and reported in only 1 case. CONCLUSIONS: Theophylline may reduce the incidence of CIN with an efficacy that is perhaps comparable to that reported in studies of N-acetylcysteine. However, findings are inconsistent across studies. A large, well-designed trial that incorporates the evaluation of clinically relevant outcomes is required to more adequately assess the role for theophylline in CIN prevention.

Angiography↗

PubMedDate not supplied

Portable and programmable clinical EOG diagnostic system.

Monitoring eye movements is clinically important in diagnosis of diseases of the central nervous system. Electrooculography (EOG) is one method of obtaining such records which uses skin electrodes, and utilizes the anterior posterior polarization of the eye. A new EOG diagnostic system has been developed that utilizes two off-the-shelf portable notebook computers, one projector and simple electronic hardware. It can be operated under Windows 95, 98, NT, and has significant advantages over any other similar equipment, including programmability, portability, improved safety and low cost. Especially, portability of the instrument is extremely important for acutely ill or handicapped patients. The purpose of this paper is to introduce the techniques of computer animation, data acquisition, real time analysis of measured data, and database management to implement a portable, programmable and inexpensive contacting EOG instrument. It is very convenient to replace the present expensive, inflexible and large-sized commercially available EOG instruments. A lot of interesting stimulation patterns for clinical application can be created easily in different shape, time sequence, and colour by programming in Delphi language. With the help of Winstar (a software package that is used to control I/O and interrupt functions of the computer under Windows 95, 98, NT), the I/O communication between two notebook computers and A/D interface module can be effectively programmed. In addition, the new EOG diagnostic system is battery operated and it has the advantages of low noise as well as isolation from electricity. Two kinds of EOG tests, pursuit and saccade, were performed on 20 normal subjects with this new portable and programmable instrument. Based on the test result, the performance of the new instrument is superior to the other commercially available instruments. In conclusion, we hope that it will be more convenient for doctors and researchers to do the clinical EOG diagnosis and basic medical science research by using this new creation.

Adult↗

PubMed2005

Large-scale extraction of gene regulation for model organisms in an ontological context.

This paper presents an approach using syntactosemantic rules for the extraction of relational information from biomedical abstracts. The results show that by overcoming the hurdle of technical terminology, high precision results can be achieved. From abstracts related to baker's yeast, we manage to extract a regulatory network comprised of 441 pairwise relations from 58,664 abstracts with an accuracy of 83 - 90%. To achieve this, we made use of a resource of gene/protein names considerably larger than those used in most other biology related information extraction approaches. This list of names was included in the lexicon of our retrained partof- speech tagger for use on molecular biology abstracts. For the domain in question an accuracy of 93.6 - 97.7% was attained on Part-of-speech-tags. The method can be easily adapted to other organisms than yeast, allowing us to extract many more biologically relevant relations. The main reason for the comparable precision rates is the ontological model that was built beforehand and served as a guiding force for the manual coding of the syntactosemantic rules.

Abstracting and Indexing↗

PubMed2005

Reading depends on writing, in Chinese.

Language development entails four fundamental and interactive abilities: listening, speaking, reading, and writing. Over the past four decades, a large body of evidence has indicated that reading acquisition is strongly associated with a child's listening skills, particularly the child's sensitivity to phonological structures of spoken language. Furthermore, it has been hypothesized that the close relationship between reading and listening is manifested universally across languages and that behavioral remediation using strategies addressing phonological awareness alleviates reading difficulties in dyslexics. The prevailing view of the central role of phonological awareness in reading development is largely based on studies using Western (alphabetic) languages, which are based on phonology. The Chinese language provides a unique medium for testing this notion, because logographic characters in Chinese are based on meaning rather than phonology. Here we show that the ability to read Chinese is strongly related to a child's writing skills and that the relationship between phonological awareness and Chinese reading is much weaker than that in reports regarding alphabetic languages. We propose that the role of logograph writing in reading development is mediated by two possibly interacting mechanisms. The first is orthographic awareness, which facilitates the development of coherent, effective links among visual symbols, phonology, and semantics; the second involves the establishment of motor programs that lead to the formation of long-term motor memories of Chinese characters. These findings yield a unique insight into how cognitive systems responsible for reading development and reading disability interact, and they challenge the prominent phonological awareness view.

Asian People↗

PubMed2005

High-recall protein entity recognition using a dictionary.

SUMMARY: Protein name extraction is an important step in mining biological literature. We describe two new methods for this task: semiCRFs and dictionary HMMs. SemiCRFs are a recently-proposed extension to conditional random fields (CRFs) that enables more effective use of dictionary information as features. Dictionary HMMs are a technique in which a dictionary is converted to a large HMM that recognizes phrases from the dictionary, as well as variations of these phrases. Standard training methods for HMMs can be used to learn which variants should be recognized. We compared the performance of our new approaches with that of Maximum Entropy (MaxEnt) and normal CRFs on three datasets, and improvement was obtained for all four methods over the best published results for two of the datasets. CRFs and semiCRFs achieved the highest overall performance according to the widely-used F-measure, while the dictionary HMMs performed the best at finding entities that actually appear in the dictionary-the measure of most interest in our intended application. AVAILABILITY: Dictionary HMMs were implemented in Java. Algorithms are available through an information extraction package MINORTHIRD on http://minorthird.sourceforge.net

Algorithms↗

PubMed2006

Machine learning and word sense disambiguation in the biomedical domain: design and evaluation issues.

BACKGROUND: Word sense disambiguation (WSD) is critical in the biomedical domain for improving the precision of natural language processing (NLP), text mining, and information retrieval systems because ambiguous words negatively impact accurate access to literature containing biomolecular entities, such as genes, proteins, cells, diseases, and other important entities. Automated techniques have been developed that address the WSD problem for a number of text processing situations, but the problem is still a challenging one. Supervised WSD machine learning (ML) methods have been applied in the biomedical domain and have shown promising results, but the results typically incorporate a number of confounding factors, and it is problematic to truly understand the effectiveness and generalizability of the methods because these factors interact with each other and affect the final results. Thus, there is a need to explicitly address the factors and to systematically quantify their effects on performance. RESULTS: Experiments were designed to measure the effect of "sample size" (i.e. size of the datasets), "sense distribution" (i.e. the distribution of the different meanings of the ambiguous word) and "degree of difficulty" (i.e. the measure of the distances between the meanings of the senses of an ambiguous word) on the performance of WSD classifiers. Support Vector Machine (SVM) classifiers were applied to an automatically generated data set containing four ambiguous biomedical abbreviations: BPD, BSA, PCA, and RSV, which were chosen because of varying degrees of differences in their respective senses. Results showed that: 1) increasing the sample size generally reduced the error rate, but this was limited mainly to well-separated senses (i.e. cases where the distances between the senses were large); in difficult cases an unusually large increase in sample size was needed to increase performance slightly, which was impractical, 2) the sense distribution did not have an effect on performance when the senses were separable, 3) when there was a majority sense of over 90%, the WSD classifier was not better than use of the simple majority sense, 4) error rates were proportional to the similarity of senses, and 5) there was no statistical difference between results when using a 5-fold or 10-fold cross-validation method. Other issues that impact performance are also enumerated. CONCLUSION: Several different independent aspects affect performance when using ML techniques for WSD. We found that combining them into one single result obscures understanding of the underlying methods. Although we studied only four abbreviations, we utilized a well-established statistical method that guarantees the results are likely to be generalizable for abbreviations with similar characteristics. The results of our experiments show that in order to understand the performance of these ML methods it is critical that papers report on the baseline performance, the distribution and sample size of the senses in the datasets, and the standard deviation or confidence intervals. In addition, papers should also characterize the difficulty of the WSD task, the WSD situations addressed and not addressed, as well as the ML methods and features used. This should lead to an improved understanding of the generalizablility and the limitations of the methodology.

Algorithms↗

PubMed1995

Biological relationships among the Iroquois.

The intrusive cultural and linguistic position of Iroquoian-speaking peoples among Algonquian speakers of the lower Great Lakes region has sparked much debate concerning their cultural and biological origins. Previous physical anthropological studies, largely limited to typological or univariate osteometric analyses, have produced equivocal and contradictory results, offering little insight into the question of origins. This study addresses the question of Iroquois origins by examining biological relationships in the lower Great Lakes region using a multivariate analysis of anthropometric data. The study uses 12 body and head measurements taken on 843 individuals from 7 Iroquoian and 4 Algonquian groups. The data were examined using biodistances, canonical variate analysis, and a model of differential gene flow. The results indicate a possible biological distinction between Iroquoians and Algonquians, suggesting different biological origins. Gene flow in prehistoric and historical times may have mitigated much of this differentiation. The Onondaga most closely approximate the ancestral Iroquoian phenotype and are phenotypic outliers, probably because of reproductive isolation from the Algonquians. The Onondaga have unexpectedly high mean phenotypic variance with respect to distance from the phenotypic centroid, suggesting a greater than average degree of external gene flow. Because no viable source for external genes can be determined, however, the Onondaga's high mean variance may be better explained through patterns of nonrandom mating and reproductive isolation.

Anthropometry↗

PubMed2006

Economic disparities in middle childhood development: does income matter?

A large literature has documented the influence of family economic resources on child development, yet income's effects in middle childhood have been understudied. Using data from the National Longitudinal Survey of Youth (N = 3,551), the author examined the influence of family income in early and middle childhood on academic skills and behavior problems during middle childhood. Early childhood income had enduring effects on children's behavior problems and academic skills in middle childhood. Middle childhood income did not influence academic skills but did affect the development of behavior problems during middle childhood. Children from low-income households were particularly sensitive to the effects of family income. The quality of home environment during early and middle childhood explained a portion of the effects of income on academic skills and behavior problems.

Achievement↗

PubMed2005

Predicting functional sites with an automated algorithm suitable for heterogeneous datasets.

BACKGROUND: In a previous report (La et al., Proteins, 2005), we have demonstrated that the identification of phylogenetic motifs, protein sequence fragments conserving the overall familial phylogeny, represent a promising approach for sequence/function annotation. Across a structurally and functionally heterogeneous dataset, phylogenetic motifs have been demonstrated to correspond to a wide variety of functional site archetypes, including those defined by surface loops, active site clefts, and less exposed regions. However, in our original demonstration of the technique, phylogenetic motif identification is dependent upon a manually determined similarity threshold, prohibiting large-scale application of the technique. RESULTS: In this report, we present an algorithmic approach that determines thresholds without human subjectivity. The approach relies on significant raw data preprocessing to improve signal detection. Subsequently, Partition Around Medoids Clustering (PAMC) of the similarity scores assesses sequence fragments where functional annotation remains in question. The accuracy of the approach is confirmed through comparisons to our previous (manual) results and structural analyses. Triosephosphate isomerase and arginyl-tRNA synthetase are discussed as exemplar cases. A quantitative functional site prediction assessment algorithm indicates that the phylogenetic motif predictions, which require sequence information only, are nearly as good as those from evolutionary trace methods that do incorporate structure. CONCLUSION: The automated threshold detection algorithm has been incorporated into MINER, our web-based phylogenetic motif identification server. MINER is freely available on the web at http://www.pmap.csupomona.edu/MINER/. Pre-calculated functional site predictions of the COG database and an implementation of the threshold detection algorithm, in the R statistical language, can also be accessed at the website.

Algorithms↗

PubMed1997

Similarities inferred from the studies of long range correlations among mitochondrial DNA sequences.

Existence of long range correlations within the DNA sequences of living organism has immense importance in understanding the language of DNA sequences. Recently it has been reported that long range correlations occur in DNA sequences. Some investigators claimed that these type of correlations occur only on intron containing DNA sequences. Some observers, however, have the opinion that long range correlations do not distinguish between the intron containing DNA sequences and intronless DNA sequences. The biological origin of long range correlations in the DNA sequences is not clearly known. In this paper we have demonstrated that long range correlations also occur on intronless mitochondrial DNA sequences, indicating that these special type of correlations are not the unique features for intron containing DNA sequences. We have also demonstrated that long range correlations simply originate in the region around which there is a large variation of pyrimidine and purine ratios. The similarities among the mitochondrial DNA sequences can be inferred by computing the fractal exponents in the region where there is a large variation of pyrimidine and purine ratio, as well as in the region where the ratio of pyrimidine and purine fluctuates in a nearly constant manner. In other words the similarities among the mitochondrial DNA sequences cannot be inferred by calculating the fractral exponents for the whole sequence.

Animals↗

PubMed2005

Examination of the effect of the polymorphic CGG repeat in the FMR1 gene on cognitive performance.

A CGG repeat sequence located in the 5' untranslated region of the FMR1 gene is polymorphic with respect to size and stability of the repeat during parent-offspring transmission. When expanded to over 200 repeats, the gene is hypermethylated and silenced, leading to fragile X syndrome (FXS). Recently, alleles with large unmethylated repeat tracts (premutations) have been associated with ovarian failure and a late-onset tremor/ataxia syndrome, symptoms unrelated to FXS. To further investigate the phenotype consequence of high repeat alleles, we have analyzed Wechsler adult intelligence scales-III (WAIS-III) measures on 66 males and 217 females with a wide range of repeat sizes. Among females only, we found that FMR1 repeat size and transcript level significantly explained approximately 4% of the variance in the Verbal IQ summary measure, suggesting that this polymorphism is one of many factors that influence variation in cognitive performance. Because of the well established association of increasing repeat size with decreasing age at menopause, we also investigated the reproductive stage and use of hormone replacement therapy (HRT) as a covariate to model verbal intelligence quotient (VIQ). We found that it explained an additional 5% of the variance in VIQ, but did not interact with FMR1 repeat and transcript level.

Adolescent↗

PubMed2003

Using rough sets, neural networks, and logistic regression to predict compliance with cholesterol guidelines goals in patients with coronary artery disease.

Coronary artery disease is a leading cause of death and disability in the United States and throughout the developed world. Results from large randomized, blinded, placebo-controlled trials have demonstrated clearly the benefit of lowering LDL cholesterol in lowering the risk for coronary artery disease. Unfortunately, despite the quantity of evidence, and the availability of medications that can efficiently lower LDL cholesterol with few side effects, not everyone who could benefit from cholesterol lowering interventions actually receives them. Despite the dissemination of national care guidelines for the evaluation and treatment of cholesterol levels (NCEP - National Cholesterol Education Program), compliance with such guidelines is suboptimal. There clearly is room for improvement in narrowing the gap between evidence based guidelines and actual clinical practice. The ability to classify those patients who are or will likely to be noncompliant on the basis of patient data routinely collected during patient care could be potentially useful by enabling the focusing of limited health care resources to those who are or will be at high risk of being under treated. In order to explore this possibility further, we attempted to create such classifiers of cholesterol guideline compliance. To do this, we obtained data from an ambulatory electronic medical record system at use at the MGH adult primary care practices for over 20 years. We obtained the data from this hierarchically-structured EMR using its own native query language, called MQL (Medical Query Language). Next, we applied to the collected data the machine learning techniques of rough set theory, neural networks (feed forward backpropagation nets), and logistic regression. We did this by using commonly available software that for the most part is freely available via the internet. We then compared the accuracy of the classifier models using the receiver operating characteristic (ROC) area and C-index summary metrics.

Cholesterol↗

PubMed2006

Endoscopic balloon sphincter dilation (sphincteroplasty) versus sphincterotomy for common bile duct stones.

BACKGROUND: Endoscopic balloon dilation was introduced as an alternative to endoscopic sphincterotomy to preserve the sphincter of Oddi and avoid undesirable effects due to an incompetent sphincter. Endoscopic balloon dilation has been largely abandoned by USA endoscopists due to increased risks of pancreatitis noted in one multicentre trial, but is still practiced in parts of Asia and Europe. OBJECTIVES: To assess the beneficial and harmful effects of endoscopic balloon dilation versus endoscopic sphincterotomy in the management of common bile duct stones. SEARCH STRATEGY: We searched The Cochrane Hepato-Biliary Group Controlled Trials Register, the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library, MEDLINE, and EMBASE until January 2004. We hand searched Gastrointestinal Endoscopy (1983-2002), read through bibliographies of all included randomised clinical trials, and contacted all primary authors regarding missed randomised trials. SELECTION CRITERIA: Randomised clinical trials comparing endoscopic balloon dilation versus endoscopic sphincterotomy in removal of common bile duct stones irrespective of publication status, language, or blinding. DATA COLLECTION AND ANALYSIS: Data collection was done by two independent authors for decisions on study inclusion, data abstraction, and quality assessment. When there was a non-resolvable discrepancy, the third author made the final decision. Analysis was run with RevMan Analysis. MAIN RESULTS: Fifteen randomised trials met our inclusion criteria (1768 participants). Less than half of the trials reported adequate methods of randomisation and only two trials used blinded outcome assessment. Endoscopic balloon dilation is statistically less successful for stone removal (relative risk (RR) 0.90, 95% confidence interval (CI) 0.84 to 0.97), requires higher rates of mechanical lithotripsy (RR 1.34, 95% CI 1.08 to 1.66), and carries a higher risk of pancreatitis (RR 1.96, 95% CI 1.34 to 2.89). Conversely, endoscopic balloon dilation has statistically significant lower rates of bleeding. When a fixed-effect model is applied endoscopic balloon dilation leads to significantly less short-term infection and long-term infection. There was no statistically significant difference with regards to mortality, perforation, or total short-term complications. AUTHORS' CONCLUSIONS: Endoscopic balloon dilation is slightly less successful than endoscopic sphincterotomy in stone extraction and more risky regarding pancreatitis. However, endoscopic balloon dilation seems to have a clinical role in patients who have coagulopathy, who are at risk for infection, and possibly in those who are older.

Catheterization↗

PubMed2026

Pharmacological therapies for the prevention of fractures in men.

RATIONALE: Pharmacological therapies for fracture prevention usually target osteoporosis, a skeletal disorder characterised by compromised bone mass or quality (or both). As most participants in osteoporosis trials are women, a review of pharmacological therapies for fracture prevention in men was warranted. OBJECTIVES: To determine the benefits and harms of bisphosphonates, parathyroid (PTH) or parathyroid-related protein (PTHrP) analogues, denosumab, and romosozumab therapy for the prevention of fractures in men. SEARCH METHODS: We searched CENTRAL, MEDLINE, Embase, and two trial registries (ClinicalTrials.gov and WHO ICTRP) until 14 October 2025, with no restrictions on date or language of publication. ELIGIBILITY CRITERIA: We included randomised controlled trials that compared bisphosphonates, PTH or PTHrP analogues, denosumab, or romosozumab (alone or with calcium or vitamin D, or both) with placebo, other drugs, or non-pharmacological therapies in men aged 50 years or older. Our primary comparison was bisphosphonates versus placebo. OUTCOMES: Critical outcomes were incidence of hip fractures, symptomatic vertebral fractures, other (not hip or vertebral) fractures, disability, participants with adverse events, study withdrawals due to adverse events, and participants with serious adverse events. Our primary time point was the final time point reported in the trials. RISK OF BIAS: We used Cochrane's RoB 2 tool to assess risk of bias. SYNTHESIS METHODS: We used a random-effects model for meta-analysis employing the Mantel-Haenszel approach, and the DerSimonian and Laird method to estimate between-trial variance. We assessed the certainty of evidence using GRADE. INCLUDED STUDIES: Seventeen trials (4132 participants) met our inclusion criteria. The average age of participants ranged from 52 to 73 years. Twelve trials used a placebo comparator versus bisphosphonate (7 trials, 2548 participants), PTH or PTHrP analogues (4 trials, 569 participants), denosumab (1 trial, 240 participants), and romosozumab (1 trial, 244 participants). For the other planned comparisons, a bisphosphonate was compared to vitamin D/vitamin D analogues (2 trials, 434 participants), to calcitonin (1 trial, 32 participants), to PTH or PTHrP analogues (1 trial, 19 participants), or to another bisphosphonate (1 trial, 301 participants), and one trial compared a bisphosphonate plus calcium to calcium tablets alone (46 participants). SYNTHESIS OF RESULTS: Placebo-controlled trials were largely susceptible to bias in selection of the reported result (83%), while most trials without a placebo control were also susceptible to bias arising from the randomisation process (100%) and in measurement of the outcome (80%). We are very uncertain about the effect of bisphosphonates on the incidence of hip fractures, symptomatic vertebral fractures, or other (non-hip non-vertebral) fractures compared to placebo at the final follow-up (up to two years). We downgraded the certainty of evidence once for risk of bias, twice for imprecision (very low event rates), and once for suspected publication bias. The certainty of evidence for incidence of other fractures was further downgraded for indirectness, as it was unclear if hip fractures were also included in the outcome. At up to two years, 2/875 participants (2 per 1000) in the bisphosphonate group reported hip fractures compared with 2/760 (3 per 1000) in the placebo group (risk ratio (RR) 0.73, 95% confidence interval (CI) 0.06 to 8.51; I² = 36%; 4 trials, 1635 participants); 5/1021 (4/1000) participants in the bisphosphonate group had a symptomatic vertebral fracture compared to 7/855 (8/1000) participants in the placebo group (RR 0.49, 95% CI 0.14 to 1.74; I² = 0%; 5 trials, 1876 participants); 25/1130 participants (16/1000) in the bisphosphonate group reported other (non-hip non-vertebral) fractures compared to 19/913 participants (21/1000) in the placebo group (RR 0.78, 95% CI 0.42 to 1.45; I² = 0%; 6 trials, 2043 participants). Bisphosphonates probably do not increase the risk of adverse events: 1024/1374 participants (746/1000) receiving bisphosphonates reported adverse events compared to 826/1174 participants (704/1000) receiving placebo (RR 1.06, 95% CI 0.93 to 1.19; I² = 75%; 7 trials, 2548 participants; moderate-certainty evidence) or serious adverse events: 329/1329 participants (272/1000) receiving bisphosphonate reported serious adverse events compared to 323/1128 participants (286/1000) receiving placebo (RR 0.95, 95% CI 0.84 to 1.08; I² = 0%; 6 trials, 2457 participants; moderate-certainty evidence). We downgraded the certainty of evidence once due to potential bias for adverse events and serious adverse events. We are very uncertain if bisphosphonates result in more withdrawals due to adverse events: 41/1374 participants (25/1000) in the bisphosphonate group withdrew due to adverse events compared with 43/1174 participants (37/1000) in the placebo group (RR 0.68, 95% CI 0.39 to 1.18; I² = 37%; 7 trials, 2548 participants; very low-certainty evidence). We downgraded the certainty of evidence once for risk of bias, once for indirectness, and once for imprecision. No trial reported disability. We are very uncertain about the effects of PTH or PTHrP analogues, denosumab, or romosozumab compared to placebo on fracture outcomes. We are very uncertain about the effects of PTH/PTHrP analogues on total adverse events, withdrawals due to adverse events, and serious adverse events. Denosumab may not increase the risk of adverse events or serious adverse events compared to placebo, while the evidence for withdrawals due to adverse events is very uncertain. Romosozumab probably does not increase the risk of adverse events and may not increase the risk of serious adverse events or result in more withdrawals due to adverse events. AUTHORS' CONCLUSIONS: We are very uncertain about the effects of bisphosphonates compared to placebo on the incidence of hip fractures, symptomatic vertebral fractures, or other (non-hip non-vertebral) fractures in men at up to two years of use. Bisphosphonates probably do not increase the risk of adverse events or serious adverse events, and we are very uncertain if they result in more withdrawals due to adverse events. We downgraded the certainty of evidence for indirectness, imprecision (low event rate), and serious risk of bias in selection of the reported result, as it was unclear if all studies fully reported every fracture. We found similar results for PTH or PTHrP analogues, denosumab, or romosozumab versus placebo. Larger, longer placebo-controlled studies are needed to determine whether pharmacological therapies are beneficial for reducing fractures in men. FUNDING: This Cochrane review had no dedicated funding. REGISTRATION: Protocol (2021): https://doi.org/10.1002/14651858.CD014707.

Humans↗

PubMed1998

The acquisition of plural marking in English and German revisited: schemata versus rules.

This article contributes to a debate in the linguistic and psychological literature that centres around the representation of morphologically complex words in the grammar and in the lexicon. The issue is whether inflectional morphology is rule-based (i.e. symbolically represented), or whether the assumption of pattern association is more adequate to account for the facts. On the basis of the analysis of acquisitional data the article strongly argues for the latter alternative. In a classic experiment that helped shape the development of acquisition theory Berko (1958) reported substantial support for ITEM-AND-PROCESS rules in the acquisition of plural morphology in English. A large part of her results were zero responses (repetition of the stimulus). A reinterpretation of these zero responses in light of schema theory and the cue strength hypothesis shows a striking departure from randomness. Berko's subjects tended to repeat stimuli just to the extent that these already resembled a plural schema. A reinterpretation of data reported in Innes (1974) achieved compatible results. This data set is far more extensive than Berko's and is used in the present study to put the schema model to a more stringent test. A reinterpretation of a parallel experiment with German children, using the cue strength analysis of the more complex plural morphology of German yielded parallel results. Finally, natural acquisitional data obtained from seven German speaking children aged between 2;1 and 2;9 are analysed. Again, strong support is found for the schema model. It is suggested that a schema-learning mechanism may underlie the acquisition of morphology, even when the end product of the learning process involves item-and-process rules, as in the case of English plural formation. In a schema-learning model, the child builds schematic representations for possible singular and plural lexical items as whole gestalts, and attempts to map concrete forms onto these schemata in deciding whether the forms have singular or plural value.

Child↗