An official website of the United States government
Abstract Accurate crack detection in concrete and pavement images is critical for infrastructure assessment but is limited by the scarcity of large, consistently annotated datasets. Supervised learning methods are particularly sensitive to data scarcity, often overfitting and generalizing poorly across crack types and imaging conditions. This study proposes a synthetic crack generation framework to augment or partially replace real datasets while reducing annotation effort. Synthetic cracks are generated by tracing minimal and maximal cumulative energy paths on random noise fields using dynamic programming, producing realistic one-pixel-wide crack strands. These are expanded via variable-width morphological dilation and deformed through geometric transformations and elastic deformation to model variations in width, tortuosity, and boundary irregularities across longitudinal, transverse, and shear cracks. The synthetic data trains a filter-based segmentation and connected component classification system rather than an end-to-end model. Over 2.25 million unique samples are generated across diverse scales and geometries. Elastic deformation increases geometric diversity, raising the mean normalized pairwise feature distance from approximately 0.17 to 0.31. Evaluation on Cracks-200, CDLN, and DeepCrack datasets shows performance comparable to human-annotated training data, with F1-scores up to 0.79 and mIoU exceeding 0.80. These results demonstrate that synthetic crack data can effectively supplement or substitute real annotated datasets, reducing annotation effort while preserving segmentation performance.
more »
« less
This content will become publicly available on December 1, 2026
Improving medical machine learning models with generative balancing for equity and excellence
Abstract Applying machine learning to clinical outcome prediction is challenging due to imbalanced datasets and sensitive tasks that contain rare yet critical outcomes and where equitable treatment across diverse patient groups is essential. Despite attempts, biases in predictions persist, driven by disparities in representation and exacerbated by the scarcity of positive labels, perpetuating health inequities. This paper introduces , a synthetic data generation approach leveraging large language models, to address these issues. enhances algorithmic performance and reduces bias by creating realistic, anonymous synthetic patient data that improves representation and augments dataset patterns while preserving privacy. Through experiments on multiple datasets, we demonstrate that boosts mortality prediction performance across diverse subgroups, achieving up to a 21% improvement in F1 Score without requiring additional data or altering downstream training pipelines. Furthermore, consistently reduces subgroup performance gaps, as shown by universal improvements in performance and fairness metrics across four experimental setups.
more »
« less
- Award ID(s):
- 2205289
- PAR ID:
- 10583752
- Publisher / Repository:
- Springer
- Date Published:
- Journal Name:
- npj Digital Medicine
- Volume:
- 8
- Issue:
- 1
- ISSN:
- 2398-6352
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
AI agents promise to empower biomedical discovery, but realizing this promise requires the ability to complete transparent, long-horizon analyses using tools. Agents must make intermediate decisions explicit, and validate each decision and output against data and tool constraints as the analysis unfolds. We present Medea, an AI agent that takes an omics objective and executes a transparent multi-step analysis using tools. Medeacomprises four modules: research planning with context and integrity verification, code execution with pre- and post-run checks, literature reasoning with evidence-strength assessment, and a consensus stage that reconciles evidence across datasets, tools, and literature. Medeauses 20 tools spanning single-cell and bulk transcriptomic datasets, cancer vulnerability maps, pathway knowledge bases, and machine learning models. We evaluate Medeaacross 5,679 analyses in three open-ended domains: target identification across five diseases and cell type contexts (2,400 analyses), synthetic lethality reasoning in seven cell lines (2,385 analyses), and immunotherapy response prediction in bladder cancer (894 patient analyses). In evaluations that vary large language models, tool sets, omics objectives, and agentic modules, Medeaimproves the performance of existing approaches by up to 46% for target identification, 22% for synthetic lethality, and 24% for immunotherapy response prediction, while maintaining low failure rates and calibrated abstention. Medeashows that verification-aware AI agents improve performance by producing transparent analyses, not simply more efficient workflows.more » « less
-
Kelso, Janet (Ed.)Abstract MotivationRecent advancements in natural language processing have highlighted the effectiveness of global contextualized representations from protein language models (pLMs) in numerous downstream tasks. Nonetheless, strategies to encode the site-of-interest leveraging pLMs for per-residue prediction tasks, such as crotonylation (Kcr) prediction, remain largely uncharted. ResultsHerein, we adopt a range of approaches for utilizing pLMs by experimenting with different input sequence types (full-length protein sequence versus window sequence), assessing the implications of utilizing per-residue embedding of the site-of-interest as well as embeddings of window residues centered around it. Building upon these insights, we developed a novel residual ConvBiLSTM network designed to process window-level embeddings of the site-of-interest generated by the ProtT5-XL-UniRef50 pLM using full-length sequences as input. This model, termed T5ResConvBiLSTM, surpasses existing state-of-the-art Kcr predictors in performance across three diverse datasets. To validate our approach of utilizing full sequence-based window-level embeddings, we also delved into the interpretability of ProtT5-derived embedding tensors in two ways: firstly, by scrutinizing the attention weights obtained from the transformer’s encoder block; and secondly, by computing SHAP values for these tensors, providing a model-agnostic interpretation of the prediction results. Additionally, we enhance the latent representation of ProtT5 by incorporating two additional local representations, one derived from amino acid properties and the other from supervised embedding layer, through an intermediate fusion stacked generalization approach, using an n-mer window sequence (or, peptide/fragment). The resultant stacked model, dubbed LMCrot, exhibits a more pronounced improvement in predictive performance across the tested datasets. Availability and implementationLMCrot is publicly available at https://github.com/KCLabMTU/LMCrot.more » « less
-
Abstract Conformal prediction provides machine learning models with prediction sets that offer theoretical guarantees, but the underlying assumption of exchangeability limits its applicability to time series data. Furthermore, existing approaches struggle to handle multi-step ahead prediction tasks, where uncertainty estimates across multiple future time points are crucial. We propose JANET (JointAdaptive predictioN-regionEstimation forTime-series), a novel framework for constructing conformal prediction regions that are valid for both univariate and multivariate time series. JANET generalises the inductive conformal framework and efficiently produces joint prediction regions with controlledK-familywise error rates, enabling flexible adaptation to specific application needs. Our empirical evaluation demonstrates JANET’s superior performance in multi-step prediction tasks across diverse time series datasets, highlighting its potential for reliable and interpretable uncertainty quantification in sequential data.more » « less
-
Abstract Designing for manufacturing poses significant challenges in part due to the computation bottleneck of Computer-Aided Manufacturing (CAM) simulations. Although deep learning as an alternative offers fast inference, its performance is dependently bounded by the need for abundant training data. Representation learning, particularly through pre-training, offers promise for few-shot learning, aiding in manufacturability tasks where data can be limited. This work introduces VIRL, a Volume-Informed Representation Learning approach to pre-train a 3D geometric encoder. The pretrained model is evaluated across four manufacturability indicators obtained from CAM simulations: subtractive machining (SM) time, additive manufacturing (AM) time, residual von Mises stress, and blade collisions during Laser Power Bed Fusion process. Across all case studies, the model pre-trained by VIRL shows substantial enhancements in generalizability, as measured by R2 regression results, with improved performance on limited data and superior predictive accuracy with larger datasets. Regarding deployment strategy, case-specific phenomenon exists where finetuning VIRL-pretrained models adversely affects AM tasks with limited data but benefits SM time prediction. Moreover, the efficacy of Low-rank adaptation (LoRA), which balances between probing and finetuning, is explored. LoRA shows stable performance akin to probing with limited data, while achieving a higher upper bound than probing as data size increases, without the computational costs of finetuning. Furthermore, static normalization of manufacturing indicators consistently performs well across tasks, while dynamic normalization enhances performance when a reliable task dependent input is available.more » « less
