搜索结果：post-trained

Bacteriophages (phages) are viruses that selectively prey on bacteria. Their use in treating antimicrobial-resistant bacterial infections is steadily increasing due to the need for alternative therapies. The application of phage therapy is not without its challenges, including difficulties associated with isolating phages against a target strain, the limited infectivity of a phage, the cost and complexity of producing well-characterised phage stocks, and the emergence of phage resistance. The directed adaptation of phage to a specific bacterial target, also known as 'phage training', leverages the natural evolutionary capacity of phages and can be used to bolster their bacterial killing abilities. Phage training dates back almost as far as phage therapy itself, being used to expand the therapeutic use of phages. Numerous reports showcase the success and benefits of phage training in vitro and its potential to operate effectively within the framework of phage therapy. However, the time needed to train a given phage, followed by genotypic and phenotypic characterisation of both pre- and post-trained phages, is a major limitation. Here, we explore oversights of the phage training process and propose some considerations and solutions to help drive the field forward to enable its feasible integration into phage therapy.

Unsupervised aspect category detection aims to identify the underlying aspect categories discussed in a given sentence without any annotated labels. Recent studies typically generate pseudo-labels from review corpora and subsequently train models in a supervised manner. However, during pseudo-label generation, existing methods either fail to capture aspect discriminability within sentences or rely solely on pre-trained models derived from general corpora, which can mislead the training process and degrade model performance. To mitigate the limitations, we propose a novel framework (SRM-CSR) that integrates lexical-level aspect-relevant information and sentence-level contextual representations for generating high-quality pseudo labels. Specifically, in the lexical-level stage, we extract Aspect-Relevant Terms (ARTs) based on two properties: (1) domain specificity, measured by word frequency divergence between the review corpus and a general corpus; and (2) semantic stability, reflected by semantic consistency across different sentences. We further introduce an entropy-driven discriminability mechanism that can assign higher weights to aspect terms within sentences based on their similarity distributions with the extracted ARTs. In the sentence-level stage, we leverage Sentence-BERT to encode sentences into contextual representations and compute the similarity between seed sentences and sentences from the review corpus. Pseudo-labels are generated independently in both stages. We construct the training set based on the consistent results produced by the two stages, which is subsequently used to train a neural classifier based on a post-trained Domain Knowledge BERT. Extensive experiments on three real-world datasets demonstrate the effectiveness of the pseudo-labeling strategy in SRM-CSR, with the proposed method achieving an average improvement of 2.9 percentage points in macro-F1 over the strongest baselines.

Tele-robotic ultrasound (US) is a novel technique that might help overcome the current shortage of radiologists and poor access to radiologists and/or sonographers in remote or rural areas. Despite the promising results of this technology in the past two decades, there is still insufficient data about its advantages and limits, as well as the implementation in routine clinical practice and the learning curve for the user. The purpose of this prospective cohort-based study is to evaluate the performance of a 5G-based tele-robotic US system for abdominal and thyroid gland assessment in a cohort of healthy volunteers and outpatients, as well as assessing the learning curve and patient satisfaction. 64 participants (23 male, 41 female) were consecutively included during the recruitment period, for a total of 51 abdominal and 37 thyroid gland US studies. The mean age was 45.23 ± 18.90 years old, and the body mass index of the abdominal cohort was 22.97 ± 2.95 kg/m2. The learning curve estimated a minimum of 20 patients for abdominal tele-robotic US training, being almost non-existent in the thyroid gland cohort. All the variables showed no-statistical differences between standard US and tele-robotic US in the abdominal post-trained cohort except the visualization of the left kidney short axis and its interpolar length. Thyroid gland variables showed no statistical differences. The mean time of exploration for the tele-robotic US for abdomen and thyroid gland examinations were 18.33 ± 6.26 min and 4.64 ± 0.97 min respectively. Most participants (> 70%) felt comfortable and safe while being examined by the tele-robotic US. Tele-robotic US achieves equal performance in comparison with the standard US when evaluating abdominal structures in this cohort of patients, as well as a relatively fast learning curve and good patient satisfaction. The performance when assessing the thyroid gland is almost identical to the standard US, which makes it a strong first candidate for a future clinical implementation.

There are widespread fears that conversational artificial intelligence (AI) could soon exert unprecedented influence over human beliefs. In this work, in three large-scale experiments (N = 76,977 participants), we deployed 19 large language models (LLMs)-including some post-trained explicitly for persuasion-to evaluate their persuasiveness on 707 political issues. We then checked the factual accuracy of 466,769 resulting LLM claims. We show that the persuasive power of current and near-future AI is likely to stem more from post-training and prompting methods-which boosted persuasiveness by as much as 51 and 27%, respectively-than from personalization or increasing model scale, which had smaller effects. We further show that these methods increased persuasion by exploiting LLMs' ability to rapidly access and strategically deploy information and that, notably, where they increased AI persuasiveness, they also systematically decreased factual accuracy.

The use of patient-specific computational modeling of cardiovascular diseases has become increasingly popular to improve patient standard of care. Most simulation approaches currently utilize the finite element method (FEM), which is very well established and succeeds in producing high-fidelity results. However, it remains too slow for use in clinical settings, especially when many-query solutions are required to determine optimal therapeutic approaches. As a step toward addressing these demands, we have developed a Neural Network Finite Element (NNFE) approach that greatly accelerates simulations of soft tissue organ function. While the NNFE method utilizes conventional FEM meshes to define the problem geometry, it leverages advancements in neural network architecture design in new GPU-based software tools to solve the governing hyperelastic material PDEs. The NNFE method has recently captured physical contact between a deformable body and a frictionless symmetry plane. In the present work, we extended the NNFE approach to simulate trileaflet heart valve closure as a critical step in moving toward patient-specific applications. Our approach addressed two critical aspects of heart valve simulations: the use of 3D solid leaflet models as opposed to shell-based leaflet models and multi-body contact between the leaflets. We verified the approach by comparing displacements of NNFE simulated closure of a single heart valve leaflet against a frictionless symmetry plane with an identical simulation in tIGAr, the open-source isogeometric analysis extension of FEniCS. The average nodal displacement error was 0.020 mm (0.47% of the maximum displacement). We further evaluated our implementation by varying leaflet collagen fiber directions to mimic physiologically accurate deformation modes. Results of the approach indicated that the observed leaflet deformation patterns agreed well with previous trileaflet simulations. Significant variations in stress were observed transmurally, underscoring the need for solid elements to model leaflet geometry. Computational speed improvements produced an approximately 100-fold speedup, with the NNFE simulations of single leaflet closure taking 0.28 s while its FE counterpart took 61 s. Full trileaflet valve models with multi-body contact simulations took approximately 5 s, whereas equivalent FEM simulations take several hours. Training the full trileaflet model took approximately 16 h and was trained over the full functional range of pressure, so that training was only required once for all subsequent simulations. We conclude that the NNFE method can be successfully used to perform rapid simulations of complex 3D soft organ systems, such as the trileaflet heart valve, that involve large deformations, 3D geometries, and multi-body contact. Moreover, the ability to perform post-trained simulations in dramatically shorter time periods underscores the promise of machine learning-based computational mechanics approaches in patient-specific predictive computational models.

This study aimed to develop and evaluate an AI-driven platform, the Adaptive RAG Assistant MRI Platform (ARAMP), for assisting in the diagnosis and reporting of brain metastases using post-contrast axial T1-weighted (AX_T1+C) MRI. In this retrospective study, 2447 cancer patients who underwent MRI between 2010 and 2022 were screened. A subset of 100 randomized patients with confirmed brain metastases and 100 matched non-cancer controls were selected for evaluation. ARAMP integrates quantitative radiomic feature extraction with an adaptive Retrieval-Augmented Generation (RAG) framework based on a large language model (LLM, GPT-4o), incorporating five authoritative medical references. Three board-certified neuroradiologists and an independent LLM (Gemini 2.0 Pro) assessed ARAMP performance. Metrics of the assessment included Pre-/Post-Trained Inference Difference, Inter-Inference Agreement, and Sensitivity. Post-training, ARAMP achieved a mean Inference Similarity score of 67.45%. Inter-Inference Agreement among radiologists averaged 30.20% (p = 0.01). Sensitivity for brain metastasis detection improved from 0.84 (pre-training) to 0.98 (post-training). ARAMP also showed improved reliability in identifying brain metastases as the primary diagnosis post-RAG integration. This adaptive RAG-based framework may improve diagnostic efficiency and standardization in radiological workflows.

Determine whether data collected from a smartphone camera can be used to detect anemia in a pediatric population. HEMO-AI (Hemoglobin Easy Measurement by Optical Artificial Intelligence), a clinical study carried out from December 2020 to February 2023, recruited patients from the Pediatric Emergency Department, Pediatric Inpatient Department and Pediatric Hematology Unit of the Haemek Medical Center, Afula, Israel. A population-based sample of 823 patients aged 6 months to 18 years who had undergone a venous blood draw for a complete blood count since being admitted to the hospital were enrolled. Patients with total leukonychia, nailbed darkening or discoloration due to medication, nail clubbing, clinically indicated jaundice, subungual hematoma, nailbed lacerations, avulsion injuries, or nail polish applied on fingernails were not eligible for study recruitment. Video and images of the patients' hand placed in a collection chamber were collected using a smartphone camera. 823 samples, 531 from a 12.2 megapixel camera and 256 from a 12.2 megapixel camera, were collected. 26 samples were excluded by the study coordinator for irregularities. 97% of fingernails and 68% of skin samples were successfully identified by a post-trained machine learning model. Separate models built to detect anemia using images taken from the Pixel 3 had an average precision of 0.64 and an average recall of 0.4, whereas models built using the Pixel 6 had an average precision of 0.8 and an average recall of 0.84. Further supplementation of training data with synthetic data boosted the precision of the latter to 0.84 and the average recall to 0.87. This study lays the groundwork for the future evolution of non-invasive, pain-free, and accessible anemia screening tools tailored specifically for pediatric patients. It identifies important sample collection parameters and design, provides critical algorithms for the pre-processing of fingernail data, and reports an initial capability to detect anemia with 87% sensitivity and 84% specificity. Prospectively registered on www.clinicaltrials.gov (Identifier: NCT04573244) on 15 September 2020, prior to subject recruitment.

Computed tomography angiography (CTA) is very popular because it is characterized by rapidity and accessibility. However, CTA is inferior to digital subtraction angiography (DSA) in the diagnosis of intracranial artery stenosis or occlusion. DSA is an invasive examination, so we optimized the quality of cephalic CTA images. We used 5000 CTA images to train multi-scale residual denoising generative adversarial network (MRDGAN). And then 71 CTA images with intracranial large arterial stenosis were treated by Super-Resolution based on Generative Adversarial Network (SRGAN), Enhanced Super-Resolution based on Generative Adversarial Network (ESRGAN) and post-trained MRDGAN, respectively. Peak signal-to-noise ratio (PSNR) and structural similarity index measurement (SSIM) of the SRGAN, ESRGAN, MRDGAN and original CTA images were measured respectively. The qualities of MRDGAN and original images were visually assessed using a 4-point scale. The diagnostic coherence of digital subtraction angiography (DSA) with MRDGAN and original images was analyzed. The PSNR was significantly higher in the MRDGAN CTA images (35.96&#xa0;&#xb1;&#xa0;1.51) than in the original (31.51&#xa0;&#xb1;&#xa0;1.43), SRGAN (25.75&#xa0;&#xb1;&#xa0;1.18) and ESRGAN (30.36&#xa0;&#xb1;&#xa0;1.05) CTA images (all P&#xa0;<&#xa0;0.001). The SSIM was significantly higher in the MRDGAN CTA images (0.95&#xa0;&#xb1;&#xa0;0.02) than in the SRGAN (0.88&#xa0;&#xb1;&#xa0;0.03) and ESRGAN (0.90&#xa0;&#xb1;&#xa0;0.02) CTA images (all P&#xa0;<&#xa0;0.01). The visual assessment was significantly higher in the MRDGAN CTA images (3.52&#xa0;&#xb1;&#xa0;0.58) than in the original CTA images (2.39&#xa0;&#xb1;&#xa0;0.69) (P&#xa0;<&#xa0;0.05). The diagnostic coherence between MRDGAN and DSA (&#x3ba;&#xa0;=&#xa0;0.89) was superior to that between original images and DSA (&#x3ba;&#xa0;=&#xa0;0.62). Our MRDGAN can effectively optimize original CTA images and improve its clinical diagnostic value for intracranial large artery stenosis.

We performed a meta-analysis to evaluate the education effects on nurses' ability to care for subjects with pressure injuries. A systematic literature search up to April 2021 was carried out, and 29 studies included 5704 nurses at the start of the study; 3800 of them were experiment or post-training and 3804 were control or per-training. They were reporting relationships between the education effects on nurses' ability to care for subjects with pressure injuries. We calculated the odds ratio (OR) or the mean difference (MD) with 95% confidence intervals (CIs) to assess the education effects on nurses' ability to care for subjects with pressure injuries using the dichotomous or continuous method with a random or fixed-effect model. Experiment or post-trained nurses had significantly higher knowledge score (MD, 10.00; 95% CI, 7.61-12.39, P&#x2009;<&#x2009;.001), number of nurses with proper knowledge (OR, 20.70; 95% CI, 10.80-39.67, P&#x2009;<&#x2009;.001), practice score (MD, 12.39; 95% CI, 5.37-19.42, P&#x2009;<&#x2009;.001), and number of nurses with proper practice (OR, 3.56; 95% CI, 1.75-7.25, P&#x2009;<&#x2009;.001), attitudes score (MD, 7.46; 95% CI, 2.94-11.99, P&#x2009;<&#x2009;.001) compared with control or pertained nurses. Training may have a beneficial effect on improving the nurses' ability to care for subjects with pressure injuries, which was obvious in improving knowledge, practice, and attitudes post-training. Further studies are required to validate these findings.

This is a factorial (2 x 2 x 2) spatial memory and cholinergic parameters study in which the factors are chronic ethanol, thiamine deficiency and naivety in Morris water maze task. Both learning and retention of the spatial version of the water maze were assessed. To assess retrograde retention of spatial information, half of the rats were pre-trained on the maze before the treatment manipulations of pyrithiamine (PT)-induced thiamine deficiency and post-tested after treatment (pre-trained group). The other half of the animals was only trained after treatment to assess anterograde amnesia (post-trained group). Thiamine deficiency, associated to chronic ethanol treatment, had a significant deleterious effect on spatial memory performance of post-trained animals. The biochemical data revealed that chronic ethanol treatment reduced acetylcholinesterase (AChE) activity in the hippocampus while leaving the neocortex unchanged, whereas thiamine deficiency reduced both cortical and hippocampal AChE activity. Regarding basal and stimulated cortical acetylcholine (ACh) release, both chronic ethanol and thiamine deficiency treatments had significant main effects. Significant correlations were found between both cortical and hippocampal AChE activity and behaviour parameters for pre-trained but not for post-trained animals. Also for ACh release, the correlation found was significant only for pre-trained animals. These biochemical parameters were decreased by thiamine deficiency and chronic ethanol treatment, both in pre-trained and post-trained animals. But the correlation with the behavioural parameters was observed only for pre-trained animals, that is, those that were retrained and assessed for retrograde retention.

Sacral fractures are often difficult to diagnose on radiographs. Computed tomography (CT) and magnetic resonance imaging (MRI) can improve the detection rate but cannot always be performed. The accuracy of artificial intelligence (AI) in detecting orthopaedic fractures is now comparable with that of orthopaedic specialists. However, the ability of AI to detect sacral fractures has not been investigated, to our knowledge. We hypothesized that the ability to detect sacral fractures on radiographs could be improved by using AI, and aimed to develop an AI model to detect sacral fractures accurately on radiographs with better accuracy than that of orthopaedic surgeons. Subjects were patients with suspected pelvic fractures for whom radiographs and CT scans had been obtained. The radiographs were labeled according to sacral fracture status based on CT results. The data set was divided into a training set (2,038 images) and a test set (200 images). Eight convolutional neural network (CNN) models were trained using the training set. Post-trained models were used to evaluate their discrimination ability. The detection ability of 4 experienced orthopaedic surgeons was also measured using the same test set. The results of fracture assessment by the orthopaedic surgeons were compared with those of the 3 CNNs with the greatest area under the receiver operating characteristic curve. Among the 8 trained models, the highest areas under the curve were for InceptionV3 (0.989), Xception (0.987), and Inception ResNetV2 (0.984). The detection rate was significantly higher for these 3 CNNs than for the orthopaedic surgeons. By enhancing the processing of probabilistic tasks and the communication of their results, AI may be better able to detect sacral fractures than orthopaedic surgeons. Diagnostic Level III. See Instructions for Authors for a complete description of levels of evidence.

The effects of sprint training on the contractile properties of human muscle fibres obtained by needle biopsy were investigated. Individual fibres were mechanically skinned and activated by Ca(2+)- and Sr(2+)-buffered solutions at pH 7.1, and allocated to distinct populations on the basis of their contractile characteristics. The majority of fibres sampled pre-training could be separated into the three major fibre groups: Populations I (24/70, 34%), II (25/70, 36%) and III (18/70, 26%), which exhibited characteristics similar to those of histochemically classified type I, IIA and IIB fibres, respectively. The remainder (3/70, 4%) represented another fibre group, with intermediate characteristics. The muscle fibres were also activated by Ca2+ at a reduced pH of 6.6, to mimic the intracellular acidification that occurs during intense exercise. Lowering pH increased the threshold for contraction by Ca2+, reduced Ca2+ sensitivity, and increased the steepness of the force-pCa relationship, in all fibres sampled from the three major fibre groups. Maximum force was not significantly reduced in any fibre population. In the post-training sample, the three major fibre types were present in different proportions: Populations I (10/52, 19%), II (20/52, 38.5%) and III (11/52, 21%). Three other fibre groups sampled in low numbers exhibited contractile characteristics intermediate between Population I and Population II. Following sprint training all of the three main fibre populations exhibited higher thresholds for contraction by, and lower sensitivities to, Sr2+ but not Ca2+, compared with the fibres sampled pre-training. Maximum force was significantly lower in Population II fibres after sprint training. At pH 6.6, post-trained Population III fibres exhibited even lower Ca2+ sensitivity, with concomitant increases in the threshold for contraction and force-pCa curve steepness.

It has been concluded from studies using retrospective data and thus quasi-experimental designs that menarche may be delayed by prepubertal athletic training. Furthermore, a causal relationship between the age of initiation of training (AIT) and the age of menarche (AOM) has been proposed. To investigate the possibility that these conclusions were erroneous and based upon analytical artifact, a computer program was used to generate random and independent AOM and AIT for a population of 30,000 "athletes". The generated mean AOM (means = 13.4 yr) and mean AIT (means = 10.0 yr) were similar to those reported in recent literature. The sampling procedure was designed such that no relationship existed between AOM and AIT in these hypothetical athletes (r = 0.002). When two subgroups (pre- or post-menarcheal training) were compared, the pre-training group was found to have a significantly later AOM than the post-trained group (means = 13.9 yr vs means = 11.7 yr; P less than 0.05). Significant correlations were found for each subgroup between AOM and AIT (r = 0.46 and 0.40 pre- and post-menarcheal training, respectively), similar to values previously reported. In conclusion, the sampling procedures performed in the present study and in similar data sets result in biased estimates of the statistical parameters. This bias accounts for the reported relationship between AOM and AIT derived using this type of quasi-experimental design, and therefore it would appear appropriate to state that the age of menarche in athletes is "later" rather than "delayed".

The excitotoxin kainic acid (10 nmol/microliter) was used to produce bilateral lesions in the nucleus basalis magnocellularis (NBM) of rats which provides extensive cholinergic innervation to the cerebral cortex. The behavioral effects of physostigmine, THA (9-amino-1,2,3,4-tetrahydroacridine hydrochloride) and NIK-247 (9-amino-2,3,5,6,7,8-hexahydro-1H-cyclopenta[b]quinoline monohydrate hydrochloride) were investigated by observing locomotor activity, shock sensitivity and passive avoidance response in the NBM-lesioned rats. Evaluation of locomotor activity and shock sensitivity in the experimental animals did not reveal any sensorimotor disturbances caused by the lesions. Oral administration of 1 and 2 mg/kg physostigmine reduced the locomotor activity in the NBM-lesioned rats, while physostigmine (0.5 mg/kg), THA (1 or 3 mg/kg) and NIK-247 (1 or 3 mg/kg) had no effect on locomotor activity. Compared with the sham-operated controls, the NBM-lesioned rats exhibited a significantly lesser deficit in the retention of the passive avoidance response. THA (1 or 3 mg/kg) and NIK-247 (1 or 3 mg/kg) elicited good retention of the passive avoidance response. Rats with NBM lesions showed impaired acquisition of a passive avoidance response when trained repeatedly at 24-h intervals. Also, when post-training NBM lesions were induced, there was rapid extinction of the acquired passive avoidance response. THA or NIK-247 administered at doses of 3 mg/kg significantly increased response latencies of post-trained NBM-lesioned rats. THA or NIK-247 administered once a day in doses of 1 or 3 mg/kg p.o. produced a very significant increase of acetylcholine in the cerebral cortex of NBM-lesioned rats after the 21st administration. These finding suggest that THA and NIK-247 exert an ameliorating effect on memory disturbance induced by NBM lesions in rats.

Half a million people die every year from smoking-related issues across the United States. It is essential to identify individuals who are tobacco-dependent in order to implement preventive measures. In this study, we investigate the effectiveness of deep learning models to extract smoking status of patients from clinical progress notes. A Natural Language Processing (NLP) Pipeline was built that cleans the progress notes prior to processing by three deep neural networks: a CNN, a unidirectional LSTM, and a bidirectional LSTM. Each of these models was trained with a pre- trained or a post-trained word embedding layer. Three traditional machine learning models were also employed to compare against the neural networks. Each model has generated both binary and multi-class label classification. Our results showed that the CNN model with a pre-trained embedding layer performed the best for both binary and multi- class label classification.

Many insects use the pattern of polarized light in the sky as a navigational cue. In this study, we use this sensory ability as a source of inspiration to create a computational orientation model based on an artificial neural network (POL-ANN). After a training phase using numerically generated sky polarization patterns, stable and convergent networks are obtained. We undertook a series of verification tests using four typical but different sky conditions and showed that the post-trained networks were able to make an accurate prediction of the direction of the sun. Comparisons between the proposed models and models based on the convolutional neural network (CNN) structure revealed the merits of the bio-inspired architecture. We further investigated the accuracy of the models based on two different (locust-like, broader; Drosophila-like, narrower) visual fields of the sky. We find that the accuracy of the computations depends on the overhead visual scene, specifically that wider fields of view perform better when information about the overhead polarization pattern is missing.

Regular aerobic exercise provides beneficial effects on human health and reduces all-cause mortality. Aerobic exercise has profound metabolic effects, and specific metabolites may reflect physiological changes. We aimed to identify endogenous metabolites that distinguish the trained from the untrained state to increase the spectrum of analytes amenable for hypothesis testing and to expand understanding of putative beneficial pathways. Cross sectional laboratory repeated measures study. Exercise testing was performed in 37 healthy male participants and serum samples were obtained before and after completion of a ten-week standardized exercise program. Samples were analyzed for routine clinical parameters and for 188 endogenous metabolites by LC-MS/MS. Indicating the effectiveness of the intervention program, parameters of sport physiology were different after training. After correcting for multiple testing, serum concentrations of several metabolites differed between the trained and untrained state. Serine and glutamate decreased in response to exercise, whereas sarcosine and kynurenine increased. Phosphatidylcholines showed a mixed response in that four species increased and three decreased. However, all seven lysophosphatidylcholines and all four plasmalogens that differed between the trained and untrained state, increased. One short-chain acylcarnitine also decreased. In receiver operator characteristics analyses, sarcosine displayed the highest AUC value (0.839; 95% CI: 0.734-0.926) in discriminating the pre- from the post-trained state. Our study detected metabolites that clearly differentiate the trained from the untrained state. These metabolites may be targeted in mechanistic studies to understand underlying biochemical pathways and could serve to improve the design, monitoring and individualization of training programs.

Poor fetal growth is associated with decrements in muscle strength likely due to changes during myogenesis. We investigated the association of poor fetal growth with muscle strength, fatigue resistance, and the response to training in the isolated quadriceps femoris. Females (20.6 years) born to term but below the 10th percentile of ponderal index (PI)-for-gestational-age (LOWPI, n=14) were compared to controls (HIGHPI, n=14), before and after an 8-week training. Muscle strength was assessed as grip-strength and as the maximal isometric voluntary contraction (MVC) of the quadriceps femoris. Muscle fatigue was assessed during knee extension exercise. Body composition and the maximal oxygen consumption (VO(2)max) were also measured. Controlling for fat free mass (FFM), LOWPI versus HIGHPI women had ~11% lower grip-strength (P=0.023), 9-24% lower MVC values (P=0.042 pre-trained; P=0.020 post-trained), a higher rate of fatigue (pre- and post-training), and a diminished training response (P=0.016). Statistical control for FFM increased rather than decreased strength differences between PI groups. The PI was not associated with VO(2)max or measures of body composition. Strength and fatigue decrements strongly suggest that poor fetal growth affects the pathway of muscle force generation. This could be due to neuromotor and/or muscle morphologic changes during development e.g., fiber number, fiber type, etc. Muscle from LOWPI women may also be less responsive to training. Indirectly, results also implicate muscle as a potential mediator between poor fetal growth and adult chronic disease, given muscle's direct role in determining insulin resistance, type II diabetes, physical activity, and so forth.

This paper presents a methodology that reflected functions by reflecting the weight matrices of an artificial neural network. One of the major problems with the connectionist approach is that trained neural networks can only associate fixed sets of input-output mappings. We provide a methodology which allows the post-trained net to associate different input-output mappings. The different mappings are reflected in a horizontal axis, reflected in a vertical axis and scaling of the initial mapping. The methodology does not train the net on the different mappings but it transforms the weight matrix of the neural network. This paper describes a novel way of utilising sigma-pi neural networks. Our new methodology manipulates sigma-pi unit's weight matrices which transform the unit's output. The weights are cast in a matrix formulation, and then transformations can be performed on the weight matrix of the sigma-pi net. To test the new methodology, the following three steps were carried out on a neural network: (1) the network was trained to perform a mapping function, f; (2) the weights of the network were transformed; and (3) the network was tested to evaluate whether it performs the reflection in the vertical axis,f(ref-vert)(x) = a - f(x). This reflects the function in one dimension. A reflection transformation was used to manipulate the network's weight matrices to obtain a reflection in the vertical axis. Note that the network was not trained to perform the reflection in the vertical axis. The transformation of the weight matrix transformed the function the output performs. This article explains the theory which enables us to perform transformations of sigma-pi networks and obtain reflections of the output by reflecting the weight matrices. These transforms empower the network to perform related mapping tasks once one mapping task has been learnt. This article explains how each transformation is performed and it considers whether a set of 'standard' transformations can indeed be derived.