Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “deep transfer model”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

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

At least 145 records · Page 8Linked to original sources
PubMed2005

Charge-transfer excitations in the model superconductor HgBa(2)CuO(4+delta).

We report a Cu -edge resonant inelastic x-ray scattering (RIXS) study of charge-transfer excitations in the 2-8 eV range in the structurally simple compound HgBa(2)CuO(4+delta) at optimal doping (T(c)=96.5 K). The spectra exhibit a significant dependence on the incident photon energy which we carefully utilize to resolve a multiplet of weakly dispersive (<0.5 eV) electron-hole excitations, including a mode at 2 eV. The observation of this 2 eV excitation suggests the existence of a remnant charge-transfer gap deep in the superconducting phase. Quite generally, our results, which include additional data for the Mott insulator La(2)CuO(4), demonstrate the importance of exploring the incident photon-energy dependence of the RIXS cross section.

Journal Article↗
PubMed2026

ProMeta: a meta-learning framework for robust disease diagnosis and prediction from plasma proteomics.

MOTIVATION: The plasma proteome offers a dynamic window of human health, capturing the real-time intersections between genetics and physiology. However, the application of deep learning to proteomics is currently hindered by a reliance on large-scale labeled datasets, rendering standard models ineffective for rare or novel diseases where patient samples are inherently scarce. RESULTS: Here, we present ProMeta, a meta-learning framework designed to enable robust disease modeling under extreme data restrictions. By integrating knowledge-guided pathway encoding with bi-level meta-optimization, ProMeta projects unstructured proteomic profiles into biologically interpretable functional tokens. This architecture allows the model to learn a global initialization containing transferable biological priors from biobank-scale data, facilitating rapid adaptation to novel tasks. Through comprehensive benchmark experiments, ProMeta consistently outperformed transfer learning and traditional machine learning baselines in both disease diagnosis and prediction tasks. In the most challenging 4-shot scenarios (utilizing only 2 cases and 2 controls), the model achieved robust generalization with an average AUROC of &#x223c;0.69, representing a 24.6% relative improvement over the best-performing baseline methods. Mechanistic investigation revealed that ProMeta disentangles cases from controls in the latent space prior to task-specific adaptation, confirming the acquisition of universal biological rules rather than rote memorization. Furthermore, gradient-based interpretation identified disease-specific protein biomarkers and functional pathways consistent with known pathophysiology. Collectively, ProMeta overcomes the data-scarcity bottleneck in precision medicine, providing a scalable, interpretable framework for characterizing the full spectrum of human diseases, particularly for rare conditions lacking extensive clinical cohorts. AVAILABILITY AND IMPLEMENTATION: The source code of ProMeta is available at GitHub (https://github.com/lihan97/ProMeta).

Proteomics↗
PubMed1997

Alveolar bone loss one year following ovariectomy in sheep.

There is little information concerning the incidence of alveolar bone loss in estrogen-deficient women. Ovariectomized sheep are valid models for study of the effects of estrogen deficiency on bone metabolism. The objective of this study was to compare alveolar bone loss in control (C) and ovariectomized sheep (OVX) at 3 and 12 months following surgery. OVX animals had decreased serum levels of 17-beta-estradiol and increased serum levels of osteocalcin, IL-6, and urinary levels of deoxypyridinoline which, taken together, suggest development of osteoporosis. The mean probing depths and percentage of sites with pocket depths 4 to 6 mm and > 6 mm were significantly greater in OVX than C at each time period and in OVX were significantly greater at 12 months that at 3 months. Gingival tissue interleukin-6 (IL-6) levels (but not the number of IL-6(+) cells) were elevated adjacent to deep periodontal pockets; however, there was no significant elevation of levels of the proinflammatory cytokines IL-1 beta and IL-8 within gingiva. Taken together, the data suggest a systemic contribution for progression of periodontal disease associated with estrogen deficiency. This may involve upregulation of systemic IL-6 synthesis and transfer to gingiva in serum, resulting in enhanced IL-6 accumulation within the gingival tissues or reduced bone density allowing for a greater amount of alveolar bone loss.

Alveolar Bone Loss↗
PubMed1987

The vascular territories (angiosomes) of the body: experimental study and clinical applications.

The blood supply to the skin and underlying tissues was investigated by ink injection studies, dissection, perforator mapping and radiographic analysis of fresh cadavers and isolated limbs. The results were correlated with previous regional studies done in this department. The blood supply is shown to be a continuous three-dimensional network of vessels not only in the skin but in all tissue layers. The anatomical territory of a source artery in the skin and deep tissues was found to correspond in most cases, giving rise to the angiosome concept. Arteries follow closely the connective tissue framework of the body. The primary supply to the skin is by direct cutaneous arteries which vary in calibre, length and density in different regions. This primary supply is reinforced by numerous small indirect vessels, which are "spent" terminal branches of arteries supplying the deep tissues. An average of 374 major perforators was plotted in each subject, revealing that there are still many more potential skin flaps. Our arterial roadmap of the body provides the basis for the logical planning of incisions and flaps. The angiosomes defined the tissues available for composite transfer.

Arteries↗
PubMed1991

[3-dimensional computer-supported model and hologram evaluations].

For the three-dimensional measurement of plaster casts and holograms a self-fabricated measurement device will be introduced. This measurement device has the following specialties: 1. All measurements will be done without any subsidiary optical instruments like spectacles, microscope, binoculars or any thing else. Therefore a tire of measuring can be carried out. 2. Only two 3 mm deep holes should be ground in the base of the plaster casts as preparatory measures for the three-dimensional measurements of the plaster casts. 3. The menu- and window-orientated software gives the possibility for the collecting of data in a data bank. 4. The precision of these measurement results, even in the case of measuring holograms, are frequently better than the common way of measuring plaster casts. 5. The precision and information by the holograms are nearly the same, compared with the common plaster casts. 6. The tracing and measuring of the angles and lines by this newly measurement device guarantees the doctor a quick overview concerning the therapeutic measures. 7. The transfer of these data in the lateral-rays and in the posterior-anterior findings leads to a better diagnostic. That means the diagnostic horizon will be widened with this new method.

Computer Simulation↗
PubMed1986

Global geochemical cycles of carbon, sulfur and oxygen.

Time resolved data on the carbon isotopic composition of carbonate minerals and the sulfur isotopic composition or sulfate minerals show a strong negative correlation during the Cretaceous. Carbonate minerals are isotopically heavy during this period while sulfate minerals are isotopically light. The implication is that carbon is being transferred from the oxidized, carbonate reservoir to the reservoir of isotopically light reduced organic carbon in sedimentary rocks while sulfur is being transferred from the reservoir of isotopically light sedimentary sulfide to the oxidized, sulfate reservoir. These apparently oppositely directed changes in the oxidation state of average sedimentary carbon and sulfur are surprising because of a well-established and easy to understand correlation between the concentrations of reduced organic carbon and sulfide minerals in sedimentary rocks. Rocks rich in reduced carbon are also rich in reduced sulfur. The isotopic and concentration data can be reconciled by a model which invokes a significant flux of hydrothermal sulfide to the deep sea, at least during the Cretaceous.

Atlantic Ocean↗
PubMed1988

A tissue heat transfer model for relating dynamic skin temperature changes to physiological parameters.

A physical and mathematical model of the superficial tissues of the body is presented which takes into account tissue physiology, structure and blood supply. The model relates transient temperature changes at the skin surface to underlying physiological parameters. The analysis is based on a one-dimensional finite difference version of the bioheat equation applied to a multi-layered model of the superficial 10 mm of body tissue. Application of the model to the volar forearm predicts that under steady-state conditions skin surface temperature is maintained primarily by heat transfer from tissues below 10 mm, to a lesser extent by perfusion and to a small extent by superficial tissue metabolism. Model predictions of the reheat curve following a 15 s cold challenge to the skin agree closely with preliminary experimental data provided by thermography. The model also provides a physical explanation for the shape of the skin temperature reheat curve. Calculations further suggest that transient skin surface temperature measurements can provide a better indication of dermal perfusion than static temperature measurements as the effects of variations in environmental conditions, deep tissue temperature and tissue metabolism can be reduced.

Hot Temperature↗
PubMed2026

Dual functional genomics reveals a broad and convergent landscape of asciminib resistance in BCR::ABL1.

BACKGROUND: Drug resistance is a constantly evolving challenge. The allosteric inhibitor asciminib is a novel therapy for chronic myelogenous leukemia (CML) that targets the myristoyl pocket of the BCR::ABL1 kinase. While it can overcome resistance to active-site inhibitors like imatinib, new resistance mutations to asciminib are emerging. The complete landscape of these mutations, particularly those outside the kinase domain or those arising from epistatic interactions between mutations, are not well understood. METHODS: This study employed a dual functional genomics approach in CML cell line models. A high-throughput adenosine base editing (ABE) screen was used to identify broad hotspots of asciminib resistance across the entire BCR::ABL1 protein. Deep mutational scanning (DMS) was then used to create a high-resolution map of all possible amino acid changes within these hotspots. An "edit-on-edit" screen was performed to investigate epistasis by introducing a library of mutations into a cell line that was pre-edited to incorporate the common imatinib-resistance mutation, Y253H. Finally, a novel F&#xf6;rster resonance energy transfer (FRET) biosensor was developed to measure the conformational state of BCR::ABL1 in live cells and link it to drug sensitivity. RESULTS: The screens identified 279 asciminib resistance mutations and revealed resistance hotspots distributed across the SH3, SH2, and kinase domains, in contrast to imatinib resistance, which is largely confined to the kinase domain. The study uncovered a potent epistatic interaction between a mutation in the SH3 domain (V73A) and a mutation in the kinase domain P-loop (Y253H), which synergistically conferred high-level resistance. The FRET biosensor demonstrated that asciminib resistance mutations tend to destabilize the "closed" inactive conformation of the ABL1 kinase. CONCLUSIONS: The landscape of asciminib resistance is broader and more complex than previously appreciated, involving mutations across multiple domains that disrupt ABL1 autoinhibition. Epistasis between mutations acquired during sequential therapies can create unexpected and potent resistance. However, these diverse genetic resistance mechanisms converge on a single biophysical measurement of the openness of the active ABL1 conformation. This provides a unified framework for understanding asciminib resistance and underscores the need for routine clinical resistance monitoring to include the SH3 and SH2 domains in first line and later line therapy.

Fusion Proteins, bcr-abl↗
PubMed2000

A three-dimensional model of lanosterol 14alpha-demethylase of Candida albicans and its interaction with azole antifungals.

The three-dimensional structure of lanosterol 14alpha-demethylase (P450(14DM), CYP51) of Candida albicans was modeled on the basis of crystallographic coordinates of four prokaryotic P450s: P450BM3, P450cam, P450terp, and P450eryF. The P450(14DM) sequence was aligned to those of known proteins using a knowledge-based alignment method. The main chain coordinates of the core regions were transferred directly from the corresponding coordinates of P450BM3. The side chain conformations of the core regions were determined by the conformations of the equivalent residues with the highest homologous scores in four crystal structures. The model was then refined using molecular mechanics and molecular dynamics. The reliability of the resulting model was assessed by Ramachandran plots, Profile-3D, hydropathy plot analysis, and by analyzing the consistency of the model with the experimental data. The structurally and functionally important residues such as the heme binding residues, the residues interacting with redox-partner protein and/or involved in electron transfer, the residues lining substrate access channel, and the substrate binding residues were identified from the model. These residues are candidates for further site-directed mutagenesis and site-specific antipeptide antibody binding experiments. The active analogue approach was employed to search the pharmacophoric conformations for 14 azole antifungals. The resulting bioactive conformations were docked into the active site of lanosterol 14alpha-demethylase of Candida albicans. All 14 azole antifungals are shown to have a similar docking mode in the active site. The halogenated phenyl group of azole inhibitors is deep in the same hydrophobic binding cleft as the 17-alkyl chain of substrate. The pi-pi stacking interaction might exist between halogenated phenyl ring of inhibitors and the aromatic ring of residue Y132. The long side chains of some inhibitors such as itraconazole and ketoconazole surpass the active site and interact with the residues in the substrate access channel. To compare with mammalian enzymes, structurally selective residues of the active site of fungal lanosterol 14alpha-demethylase are distributed in the C terminus of F helix, beta6-1 sheet and beta6-2 sheet.

Amino Acid Sequence↗
PubMed2005

Illumination intensity dependence of the photovoltage in nanostructured TiO2 dye-sensitized solar cells.

The open-circuit voltage (V(oc)) dependence on the illumination intensity (phi0) under steady-state conditions in both bare and coated (blocked) nanostructured TiO2 dye-sensitized solar cells (DSSCs) is analyzed. This analysis is based on a recently reported model [Bisquert, J.; Zaban, A.; Salvador, P. J. Phys. Chem. B 2002, 106, 8774] which describes the rate of interfacial electron transfer from the conduction band of TiO2 to acceptor electrolyte levels (recombination). The model involves two possible mechanisms: (1) direct, isoenergetic electron injection from the conduction band and (2) a two-step process involving inelastic electron trapping by band-gap surface states and subsequent isoenergetic transfer of trapped electrons to electrolyte levels. By considering the variation of V(oc) over a wide range of illumination intensities (10(10) < phi0 < 10(16) cm(-2) s(-1)), three major regions with different values of dV(oc)/d phi0 can be distinguished and interpreted. At the lower illumination intensities, recombination mainly involves localized band-gap, deep traps at about 0.6 eV below the conduction band edge; at intermediate photon fluxes, recombination is apparently controlled by a tail of shallow traps, while, for high enough phi0 values, conduction band states control the recombination process. The high phi0 region is characterized by a slope of dV(oc)/d log phi0 congruent with 60 mV, which indicates a recombination of first order in the free electron concentration. The study, which was extended to different solar cells, shows that the energy of the deep traps seems to be an intrinsic property of the nanostructured TiO2 material, while their concentration and also the density ([symbol: see text]t approximately 10(18)-10(19) cm(-3)) and distribution of shallow traps, which strongly affects the shape of the V(oc) vs phi0 curves, change from sample to sample and are quite sensitive to the electrode preparation. The influence of the back-reaction of electrons from the fluorine-doped tin oxide (FTO) conducting glass substrate with electrolyte tri-iodide ions on the V(oc) vs phi0 dependence characteristic of the DSSC is analyzed. It is concluded that this back-reaction route can be neglected, even at low light intensities, when its rate (exchange current density, j0), which can vary over 4 orders of magnitude depending on the type of FTO used, is low enough (j0 < or = 10(-8)A cm(-2)). The comparison of V(oc) vs phi0 measurements corresponding to different DSSCs with and without blocking of the FTO-electrolyte contact supports this conclusion.

Journal Article↗
PubMed2001

Hospital in the nursing home. Treating acute hospital problems in nursing home residents using a Hospital in the Home model.

AIM: Residents of nursing homes are frequently transferred to hospital for the treatment of episodes of acute illness, in particular infectious diseases requiring intravenous therapy. This paper describes the ability of a Hospital in the Home Unit (HHU) to manage acute episodes while the patient remains resident in their nursing home. METHOD: Retrospective descriptive survey of all nursing home patients treated in one HHU from January 1996 until May 2001 (65 months). Relevant demographic and clinical outcome measures were extracted from the HHU database. RESULTS: Twenty-eight patients with a mean age of 81 years were studied. Most patients were resident in a nursing home for dementia. The commonest conditions treated were pneumonia, deep venous thrombosis, renal tract infection and cellulitis. Outcome of treatment was satisfactory in each case. No mortality was recorded. DISCUSSION: It may be feasible and safe to treat residents of nursing homes with acute medical conditions in their usual place of residence. A hospital based HHU is well placed to deliver acute medical and nursing care to such residents managing intravenous therapy and lines, allowing the nursing home to continue to manage the resident without physical transfer to hospital.

Acute Disease↗
PubMed2000

Efficiency of adenovirus-mediated gene transfer to oropharyngeal epithelial cells correlates with cellular differentiation and human coxsackie and adenovirus receptor expression.

Adenovirus-mediated gene transfer is a novel treatment strategy for head and neck squamous cell carcinoma (HNSCC) that may improve the unacceptable morbidity and mortality associated with conventional treatment. Efficient adenoviral (AdV) infection largely depends on cellular expression of the human coxsackie and adenovirus receptor (hCAR); however, the relatively recent identification of this receptor precludes a comprehensive description of its tissue distribution. We have created tissue culture model systems that approximate the differentiation and three-dimensional structure of stratified squamous epithelium characteristic of head and neck mucosa. Using these systems, we have found that expression of hCAR in native and modeled normal oropharyngeal epithelium decreased as cells differentiated with the most superficial and differentiated cells expressing no detectable hCAR. In contrast, modeled stratified HNSCC cells, which did not differentiate morphologically and did not express cytokeratin markers of differentiation, had equivalent expression of hCAR in superficial and basal layers. The expression of hCAR in our models correlated not only with the undifferentiated state, but also with efficiency of AdV infection. Despite expression of hCAR in underlying basal and suprabasal cells, topical application of AdV to normal modeled epithelium resulted in inefficient transduction of the most superficial cell layer without any infection of underlying cells. These data suggest that in normal epithelium the overlying squamous cells act as a barrier preventing infection of underlying cells that would otherwise be easily infected. In modeled stratified HNSCC, transduction was much more efficient and occurred up to four cell layers deep, suggesting that unlike normal superficial epithelial cells, the superficial cells of stratified HNSCC do not act as an effective barrier to adenoviral infection. The distribution of hCAR in native tissue and the enhanced susceptibility of undifferentiated oropharyngeal epithelial cells, including undifferentiated cancer cells, to AdV infection has important implications for the development of AdV-based targeting strategies for the treatment of head and neck cancer or premalignancies.

Adenoviridae↗
PubMed2004

Prefabrication of bone by vascular induction: an experimental study in rabbits.

The various methods of prefabricating flaps include vascular induction through staged transfer; pretransfer delay, expansion, and grafting; the use of alloplastic materials; and tissue bioengineering. In this experimental study, vascular induction was used to provide an axial vascular pedicle to randomly nourished tissue. Twenty-six New Zealand rabbits, ages and weights ranging between 6 months-1 year and 1.5-3.5 kg, respectively, were used. The sex difference was not considered. The deep inferior epigastric artery and vein were used to carry blood and were placed into a segment of femur to prefabricate the bone. Four weeks later, the viability of the segment of bone prefabricated by new axial pedicle was shown by scintigraphic study, and the new axial pedicle was ready for free transfer.

Animals↗
PubMed2005

Structures of a putative RNA 5-methyluridine methyltransferase, Thermus thermophilus TTHA1280, and its complex with S-adenosyl-L-homocysteine.

The Thermus thermophilus hypothetical protein TTHA1280 belongs to a family of predicted S-adenosyl-L-methionine (AdoMet) dependent RNA methyltransferases (MTases) present in many bacterial and archaeal species. Inspection of amino-acid sequence motifs common to class I Rossmann-fold-like MTases suggested a specific role as an RNA 5-methyluridine MTase. Selenomethionine (SeMet) labelled and native versions of the protein were expressed, purified and crystallized. Two crystal forms of the SeMet-labelled apoprotein were obtained: SeMet-ApoI and SeMet-ApoII. Cocrystallization of the native protein with S-adenosyl-L-homocysteine (AdoHcy) yielded a third crystal form, Native-AdoHcy. The SeMet-ApoI structure was solved by the multiple anomalous dispersion method and refined at 2.55 A resolution. The SeMet-ApoII and Native-AdoHcy structures were solved by molecular replacement and refined at 1.80 and 2.60 A, respectively. TTHA1280 formed a homodimer in the crystals and in solution. Each subunit folds into a three-domain structure composed of a small N-terminal PUA domain, a central alpha/beta-domain and a C-terminal Rossmann-fold-like MTase domain. The three domains form an overall clamp-like shape, with the putative active site facing a deep cleft. The architecture of the active site is consistent with specific recognition of uridine and catalysis of methyl transfer to the 5-carbon position. The cleft is suitable in size and charge distribution for binding single-stranded RNA.

Amino Acid Motifs↗
PubMed2025

Efficient Detection and Characterization of Targets of Natural Selection Using Transfer Learning.

Natural selection leaves detectable patterns of altered spatial diversity within genomes, and identifying affected regions is crucial for understanding species evolution. Recently, machine learning approaches applied to raw population genomic data have been developed to uncover these adaptive signatures. Convolutional neural networks (CNNs) are particularly effective for this task, as they handle large data arrays while maintaining element correlations. However, shallow CNNs may miss complex patterns due to their limited capacity, while deep CNNs can capture these patterns but require extensive data and computational power. Transfer learning addresses these challenges by utilizing a deep CNN pretrained on a large dataset as a feature extraction tool for downstream classification and evolutionary parameter prediction. This approach reduces extensive training data generation requirements and computational needs while maintaining high performance. In this study, we developed TrIdent, a tool that uses transfer learning to enhance detection of adaptive genomic regions from image representations of multilocus variation. We evaluated TrIdent across various genetic, demographic, and adaptive settings, in addition to unphased data and other confounding factors. TrIdent demonstrated improved detection of adaptive regions compared to recent methods using similar data representations. We further explored model interpretability through class activation maps and adapted TrIdent to infer selection parameters for identified adaptive candidates. Using whole-genome haplotype data from European and African populations, TrIdent effectively recapitulated known sweep candidates and identified novel cancer, and other disease-associated genes as potential sweeps.

Selection, Genetic↗
PubMed2004

Cellular parameters for track structure modeling of radiation hazard in space.

Based on irradiation with 45 MeV/u N and B ions and with Co-60 gamma rays, cellular parameters of Katz's track structure model have been fitted for the survival of V79-379A Chinese hamster lung fibroblasts. Cellular parameters representing neoplastic transformations in C3H10T/1/2 cells after their irradiation with heavy ion beams, taken from earlier work, were also used to model the radiation hazard in deep space, following the system for evaluating, summing and reporting occupational exposures proposed in 1967 by a subcommittee of NCRP. We have performed model calculations of the number of transformations in surviving cells, after a given fluence of heavy charged particles of initial energy 500 MeV/u, penetrating thick layers of cells. We take the product of cell transformation and survival probabilities, calculated along the path lengths of charged particles using cellular survival and transformation parameters, to represent a quantity proportional to the "radiation risk factor" discussed in the NCRP document. The "synergistic" effect of simultaneous charged particle transfers is accounted for by the "track overlap" mode inherent in the model of Katz.

Animals↗
PubMed2004

The ring of life provides evidence for a genome fusion origin of eukaryotes.

Genomes hold within them the record of the evolution of life on Earth. But genome fusions and horizontal gene transfer seem to have obscured sufficiently the gene sequence record such that it is difficult to reconstruct the phylogenetic tree of life. Here we determine the general outline of the tree using complete genome data from representative prokaryotes and eukaryotes and a new genome analysis method that makes it possible to reconstruct ancient genome fusions and phylogenetic trees. Our analyses indicate that the eukaryotic genome resulted from a fusion of two diverse prokaryotic genomes, and therefore at the deepest levels linking prokaryotes and eukaryotes, the tree of life is actually a ring of life. One fusion partner branches from deep within an ancient photosynthetic clade, and the other is related to the archaeal prokaryotes. The eubacterial organism is either a proteobacterium, or a member of a larger photosynthetic clade that includes the Cyanobacteria and the Proteobacteria.

Bacteria↗
PubMed2025

Efficient detection and characterization of targets of natural selection using transfer learning.

Natural selection leaves detectable patterns of altered spatial diversity within genomes, and identifying affected regions is crucial for understanding species evolution. Recently, machine learning approaches applied to raw population genomic data have been developed to uncover these adaptive signatures. Convolutional neural networks (CNNs) are particularly effective for this task, as they handle large data arrays while maintaining element correlations. However, shallow CNNs may miss complex patterns due to their limited capacity, while deep CNNs can capture these patterns but require extensive data and computational power. Transfer learning addresses these challenges by utilizing a deep CNN pre-trained on a large dataset as a feature extraction tool for downstream classification and evolutionary parameter prediction. This approach reduces extensive training data generation requirements and computational needs while maintaining high performance. In this study, we developed TrIdent, a tool that uses transfer learning to enhance detection of adaptive genomic regions from image representations of multilocus variation. We evaluated TrIdent across various genetic, demographic, and adaptive settings, in addition to unphased data and other confounding factors. TrIdent demonstrated improved detection of adaptive regions compared to recent methods using similar data representations. We further explored model interpretability through class activation maps and adapted TrIdent to infer selection parameters for identified adaptive candidates. Using whole-genome haplotype data from European and African populations, TrIdent effectively recapitulated known sweep candidates and identified novel cancer, and other disease-associated genes as potential sweeps.

Journal Article↗