Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Transfer learning”

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 37 records · Page 2Linked to original sources
PubMed2026

Listening forward: emerging roles of bioacoustics in ecology, evolution, and conservation.

Bioacoustics is increasingly shifting from a mostly descriptive pursuit to one that can anticipate ecological change. Recent innovations-from autonomous recording units and edge-computing sensors to speech-inspired feature extraction and machine-learning techniques like transfer learning, unsupervised discovery, and explainable AI-are transforming the study of animal communication. These advances let us work at scales previously difficult to imagine. Automated species recognition, individual identification, and even tracking cultural evolution over decades are now within reach. Entire ecosystem soundscapes can be mapped with unprecedented resolution. Looking ahead, global listening networks, adaptive acoustic indices, and live biodiversity dashboards seem increasingly realistic. We may soon build digital models that simulate communication networks under future scenarios. Closer integration with genomics, physiology, and robotics could link vocal traits to their genetic, physiological, and ecological drivers. Challenges remain, including data governance, acoustic privacy, and equitable access to the planet's sonic heritage. Bioacoustics may be on the way to becoming a predictive, integrative science - one particularly well suited to monitoring, interpreting, and helping safeguard life's communication systems in a rapidly changing world.

Animals↗
PubMed2023

Unraveling Neuronal Identities Using SIMS: A Deep Learning Label Transfer Tool for Single-Cell RNA Sequencing Analysis.

Large single-cell RNA datasets have contributed to unprecedented biological insight. Often, these take the form of cell atlases and serve as a reference for automating cell labeling of newly sequenced samples. Yet, classification algorithms have lacked the capacity to accurately annotate cells, particularly in complex datasets. Here we present SIMS (Scalable, Interpretable Machine Learning for Single-Cell), an end-to-end data-efficient machine learning pipeline for discrete classification of single-cell data that can be applied to new datasets with minimal coding. We benchmarked SIMS against common single-cell label transfer tools and demonstrated that it performs as well or better than state of the art algorithms. We then use SIMS to classify cells in one of the most complex tissues: the brain. We show that SIMS classifies cells of the adult cerebral cortex and hippocampus at a remarkably high accuracy. This accuracy is maintained in trans-sample label transfers of the adult human cerebral cortex. We then apply SIMS to classify cells in the developing brain and demonstrate a high level of accuracy at predicting neuronal subtypes, even in periods of fate refinement, shedding light on genetic changes affecting specific cell types across development. Finally, we apply SIMS to single cell datasets of cortical organoids to predict cell identities and unveil genetic variations between cell lines. SIMS identifies cell-line differences and misannotated cell lineages in human cortical organoids derived from different pluripotent stem cell lines. When cell types are obscured by stress signals, label transfer from primary tissue improves the accuracy of cortical organoid annotations, serving as a reliable ground truth. Altogether, we show that SIMS is a versatile and robust tool for cell-type classification from single-cell datasets.

Brain organoids↗
PubMed2026

Reliability-aware hierarchical learning for Chagas disease screening from 12-lead ECGs: tackling label uncertainty and class imbalance.

Objective.Chagas disease, a neglected tropical disease (NTD) with significant cardiovascular impact, remains underdiagnosed in resource-limited regions. Electrocardiogram (ECG) screening offers a low-cost tool for detecting cardiac involvement, yet algorithm development is challenged by label noise, data scarcity, and the latent nature of infection. This study proposes a robust ECG-based screening framework that explicitly addresses these constraints.Approach.We introduce aReliability-Aware Hierarchical Learningstrategy that calibrates supervision according to data provenance, prioritizing serology-confirmed labels over noisy self-reports. To mitigate data scarcity, we compare a specialized convolutional neural network (CNN) trained from scratch with a transfer learning approach based on a Spatio-Temporal ECG foundation Model (FM). Performance is evaluated across varying data scales, and the representation structure is analyzed to interpret model behavior.Main results.On the official hidden test set of the George B. Moody PhysioNet/Computing in Cardiology Challenge 2025, our approach achieved a Challenge Score of 0.163. We observe that while the specialized CNN performs competitively in data-rich regimes, the FM exhibits superior robustness in extreme low-resource settings. Furthermore, performance reaches a plateau imposed by underlying disease physiology. Bimodal score distributions suggest that models distinguish established cardiomyopathy from indeterminate infection, which remains electrophysiologically indistinguishable from healthy controls.Significance.These findings clarify both the potential and intrinsic limits of ECG-based AI screening for NTD-associated cardiac involvement. Reliability-aware supervision and data-efficient transfer learning provide a practical framework toward scalable and clinically meaningful ECG screening systems in resource-constrained environments.

Humans↗
PubMed1976

A behavioral manifestation of the development of interhemispheric transfer of learning in children.

Two hundred and fifty normal, right-handed children ranging from aged 5 to 15 were administered measures of grip strength, finger-tapping speed and the Tactual Performance Test. In order to assess the development of manual preference, analysis for each of the measures were performed on the left hand score as a percentage of the right hand score. Small age-related trends were found for the pure motor measures of speed and strength. In the case of the TPT, a clear age-related practice effect was found. The results were discussed in terms of the development of hemispheric cooperation.

Adolescent↗
PubMed2025

Deep learning-based annotation of plant abiotic stress resistance genes for crops.

The declining costs of DNA sequencing have expanded genomic data, crucial for understanding plant abiotic stress responses and crop improvement. However, accurate gene annotation remains challenging. To address this limitation, we propose the PASRGA, a deep learning approach that leverages transfer learning and contrastive learning to annotate genes related to drought, salt, cold, and UV resistance. PASRGA achieves high F1-scores, area under the receiver operating characteristic (AUROC), area under the precision-recall curve (AUPRC), and Matthews correlation coefficient (MCC) in annotating stress resistance genes, significantly outperforming the general protein annotation model CLEAN, the plant phosphatase gene annotation model PF-NET, the top-ranked model in the CAFA5 challenge NetGO 4.0, and four traditional machine learning methods. Its effectiveness was further validated with a salt stress treatment experiment in Eutrema salsugineum. To facilitate crop breeding practices, we utilized PASRGA to annotate the genomes of 17 major crops. To improve accessibility and utility, we incorporated both manually curated and PASRGA-predicted gene data, together with the PASRGA tool, into the PlantASRG database (https://bioinfor.nefu.edu.cn/PlantASRG/). This comprehensive resource aims to support crop breeding initiatives and ensure food security.

Crops, Agricultural↗
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 ∼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↗
PubMed2026

pKAKA: a protein language model for prioritizing kinase-disrupting variants in diseases.

Protein kinases are pivotal regulators of cellular signaling, and their genetic variations are frequently implicated in diseases. Although numerous kinase mutations have been identified as drivers of altered activity, with a few successfully targeted therapeutically, the functional impact of most variants remains uncharacterized. To bridge this gap, we curate a comprehensive dataset that contains 2553 experimentally validated kinase activity-related key alterations (KAKAs) from the literature. While many mutations outside canonical functional regions are known to affect kinase activity, systematic methods to predict their functional consequences are lacking. Consequently, we develop a computational method to predict potential KAKAs, leveraging transfer learning on the pre-trained protein language model ProtBert. Our model, termed pKAKA, achieves an impressive AUC score of 0.9593 and outperforms the AlphaMissense benchmark in comparative testing. Systematic analysis of kinase missense mutations underscores the critical role of KAKAs in pathogenesis, with highlights including JAK2 V617F in atherosclerotic cardiovascular disease, LRRK2 G2385R in Parkinson's disease, EGFR L858R in lung adenocarcinoma, and EGFR G598V in glioma. Overall, this study significantly advances our understanding of how mutations that influence kinase activity contribute to disease mechanisms.

Humans↗
PubMed2026

Predictive design of tissue-specific mammalian enhancers that function in the mouse embryo.

Enhancers control tissue-specific gene expression across animals1. Although deep learning2,3 has enabled enhancer prediction and design in mammalian cell lines and non-mammalian model organisms4-10 (reviewed in a previous publication11), it remains unclear whether such approaches can operate within the regulatory complexity of mammalian genomes and tissues in vivo. Here we present a general strategy for designing tissue-specific enhancers that function reliably in mice. We use deep learning to train compact convolutional neural networks on curated chromatin accessibility data and fine-tune them by transfer learning on validated human and mouse enhancers. Guided by these models, we design 15 synthetic enhancers for the heart, limb and central nervous system in mouse embryos, all of which are active in their intended target tissue. These results demonstrate that mammalian enhancer function can be reliably inferred from DNA sequence alone, enabling the predictive de novo design of tissue-specific synthetic enhancers from modest training sets. This work establishes a generalizable framework for programmable control of mammalian gene expression in vivo, opening new avenues in functional genomics, synthetic biology and gene therapy.

Animals↗
PubMedDate not supplied

Recovery of function after brain damage: the extrinsic stimulus specificity of spared neural traces.

Previous research has demonstrated that while primary visual cortex lesions in rats will disrupt performance of a learned brightness discrimination, the neural traces associated with this discrimination are spared. Moreover, the spared neural traces may, in certain instances, influence the postoperative reinstatement of a brightness discrimination. The present research was designed to determine the generality of this postoperative influence. The results indicate that while the normal rat and the visual decorticate rat may readily transfer learning between discriminations involving different discriminative cues, a visual lesion between two discriminations prevents this transfer. These data are interpreted as indicating that the postoperative influence of a spared neural trace is dependent upon the reoccurrence of specific extrinsic stimulus cues.

Animals↗
PubMed2026

A novel deep learning-driven framework for improving lncRNA comprehensive annotation with LncADeep 2.0.

MOTIVATION: Long non-coding RNAs (lncRNAs) have emerged as crucial players in diverse physiological and pathological processes, yet the biological mechanisms of the vast majority of lncRNAs remain elusive. To fill this gap, it is necessary to improve the accuracy of lncRNA identification and functional annotation. RESULTS: Here, we introduce LncADeep 2.0, an integrated deep learning framework designed to meet these needs. In the identification module, LncADeep 2.0 incorporated novel peptide features along with sequence and structural information, demonstrating superior performance over our previous LncADeep and other existing tools on both annotated transcripts from GENCODE and RNA-seq data. For functional annotation, LncADeep 2.0 leveraged lncRNA-centric interaction networks and gene ontology terms through the transfer learning strategy to achieve robust annotation performance with limited functional data. Compared to LncADeep, LncADeep 2.0 could accurately elucidate the general functions of given lncRNA sequences, predict tissue- or cell-type-specific functions from bulk and single-cell RNA-seq data, and establish connections between tumor-associated lncRNAs and genomic markers. Overall, LncADeep 2.0 stands out as an efficient and reliable tool for lncRNA identification and functional annotation across a wide spectrum of biological processes. AVAILABILITY AND IMPLEMENTATION: LncADeep 2.0 is available for use at https://github.com/Jefferson-Chou/LncADeep2 and https://doi.org/10.5281/zenodo.17164767.

RNA, Long Noncoding↗
PubMed2025

OpenSpliceAI: An efficient, modular implementation of SpliceAI enabling easy retraining on non-human species.

The SpliceAI deep learning system is currently one of the most accurate methods for identifying splicing signals directly from DNA sequences. However, its utility is limited by its reliance on older software frameworks and human-centric training data. Here we introduce OpenSpliceAI, a trainable, open-source version of SpliceAI implemented in PyTorch to address these challenges. OpenSpliceAI supports both training from scratch and transfer learning, enabling seamless retraining on species-specific datasets and mitigating human-centric biases. Our experiments show that it achieves faster processing speeds and lower memory usage than the original SpliceAI code, allowing large-scale analyses of extensive genomic regions on a single GPU. Additionally, OpenSpliceAI's flexible architecture makes for easier integration with established machine learning ecosystems, simplifying the development of custom splicing models for different species and applications. We demonstrate that OpenSpliceAI's output is highly concordant with SpliceAI. In silico mutagenesis (ISM) analyses confirm that both models rely on similar sequence features, and calibration experiments demonstrate similar score probability estimates.

Journal Article↗
PubMed1979

Clinical therapy for accommodative responses: transfer effects upon performance.

A clinical therapy program featuring accommodative training was administered to a group of children with diagnosed disorders of accommodative function. The children ranged in age from six to twelve years. A group of subjects representing the same clinical population, and not differing significantly in age or grade level, acted as a control group. The control subjects participated in a therapy program of a similar duration, wherein perceptual-motor training (unrelated to the training of accommodative skills) was administered. A nearpoint pencil-and-paper task was administered to all subjects before and after the training programs, to assess changes in performance as a criterion of learning transfer and behavioral generalization. A significantly greater decrease in errors occurred in the group receiving the accommodative training as contrasted to the control group. No significant differences were found in the time scores. The results suggest that accommodative training, for children with diagnosed accommodative disorders, has transfer effects upon nearpoint performance relating to improved accuracy.

Accommodation, Ocular↗
PubMed2026

Automated CEAP Classification of Venous Duplex Reports Using Multimodal Artificial Intelligence.

OBJECTIVE: To develop and internally validate a prototype multimodal artificial intelligence system for automated CEAP (Clinical, Etiological, Anatomical and Pathophysiological) classification of venous duplex ultrasound (VDUS) reports, integrating natural language processing of free-text components with computer vision analysis of hand-drawn anatomical diagrams. METHODS: Single centre retrospective observational study using routinely collected clinical data. One thousand consecutive venous duplex ultrasound reports from Cambridge University Hospitals NHS Foundation Trust, UK (July 2024 - May 2025) were labelled according to the CEAP classification, excluding the Etiological component, which could not be reliably determined from duplex reports alone. Transfer learning was applied using ClinicalBERT for text and MobileNetV3 for diagrammatic data. Clinical classes were predicted from request line text. Text- and image-based pathophysiological models were developed for four anatomical territories (Great Saphenous Vein, Small Saphenous Vein, Deep system, Perforators), combined using late fusion with probability averaging. RESULTS: The clinical CEAP model achieved accuracy of 0.91, macro-F1 of 0.82, and macro-AUC of 0.98. Pathophysiological prediction varied, with text models broadly outperforming image models. Fusion yielded heterogeneous benefits, improving SSV performance but reducing Deep system accuracy. The performance of the final pathophysiological CEAP fusion models varied across anatomical territories: accuracy ranged from 0.70-0.92 and macro-AUC from 0.80-0.92. CONCLUSION: This study demonstrates the feasibility of automated CEAP classification from VDUS reports. Despite class imbalance affecting minority class predictions, the strong discriminatory performance validates this multimodal ML model for extracting clinically meaningful information from real-world data. This approach offers potential, pending external validation, to streamline vascular services through automated triage and guideline-compliant decision making.

Artificial intelligence↗
PubMed2025

Predicting dynamic expression patterns in budding yeast with a fungal DNA language model.

Predicting gene expression from DNA sequence remains challenging due to complex regulatory codes. We introduce a masked DNA language model pretrained on 165 fungal genomes closely related to budding yeast that captures conserved regulatory grammar. Fine-tuning the LM on yeast RNA-seq data-including high-resolution transcriptional regulator induction time courses generated in this study-yielded Shorkie, a model that substantially improves gene expression prediction compared to baselines trained without self-supervision. Shorkie identified canonical transcription factor (TF) binding motifs and tracked their usage across induction experiments. Furthermore, Shorkie accurately predicted variant effects, outperforming leading sequence-to-expression models in cis-eQTL classification and achieving high concordance with massively parallel reporter assays. Interpretability analyses revealed Shorkie's ability to resolve promoter dynamics, splicing signals, and temporal changes in regulatory motif usage. This framework demonstrates that evolutionary-scale pretraining combined with transfer learning substantially improves our ability to decode gene regulation from sequence, providing insights into noncoding variants and regulatory networks.

Journal Article↗
PubMed1979

Learning and interhemispheric transfer of visual pattern discriminations following unilateral suprasylvian lesions in split-chiasm cats.

A suprasylvian lesion removing cortical areas 7 and 21 and portions of area 19 and of the lateral suprasylvian area was placed in one hemisphere of split-chiasm cats. By comparison with the normal side and with cortically intact split-chiasm and split-brain cats, form discrimination learning with the eye on the injured side was severely retarded. This deficit could not be attributed to an unintentional undercutting of areas 17 and 18, since in three cases the laminae of the lateral geniculate nucleus showed little retrograde atrophy; marked degeneration was found in the medial interlaminar nucleus and the pulvinar complex. In addition, interocular transfer of form discriminations to the eye on the injured side was absent or poor, while transfer in the opposite direction was normal. A cat with a suprasylvian lesion undercutting areas 17 and 18 was unable to learn pattern discriminations with the eye on the injured side, in spite of prolonged training with that eye and normal learning with the other eye. Another cat with a suprasylvian lesion selectively removing the anteromedial and posteromedial portions of the lateral suprasylvian area showed no learning deficit on the injured side, but poor transfer to that side. A learning deficit on the side of the lesion emerged in this cat after forebrain commissurotomy. The results support the hypothesis of a major involvement of cortical areas outside of 17 and 18 in the processes of abstraction and generalization of visual information necessary for learning and interhemispheric transfer of form discrimination in the cat.

Animals↗
PubMed1978

Performance of MA-matched nonretarded and retarded children on discrimination learning and transfer-shift tasks.

The effects of the nature of the relevant dimension, response mode, imagery, and age on discrimination learning and shift performance was examined. Three age groups were used, CA 8 years-MA 8 years, CA 5 years-MA 5 years. CA 8 years-MA 5 years. The latter group learned both intradimensional and extradimensional shifts slower than did the former two groups. Subsequent analyses suggested that the reason for the slower learning in the retarded group was their greater frequency of control by dominant dimensions and/or the initial control by novel stimuli. A second finding was related to the particular procedure employed. Children were required to either press a button to indicate which of two stimuli was correct or pick up the correct stimulus. Form-relevant but not color-relevant problems were facilitated by the pick-up procedure. These findings suggest that dimensional dominance may be a consequence of the procedures employed in previous reports where children were or were not permitted to manipulate the stimuli.

Age Factors↗