Frontiers in Artificial Intelligence最新文献

Purpose: This study aims to classify Kellgren-Lawrence (KL) osteoarthritis stages using knee anteroposterior X-ray images by comparing two deep learning (DL) methodologies: a traditional single-model approach and a proposed multi-model approach. We addressed three core research questions in this study: (1) How effective are single-model and multi-model deep learning approaches in classifying KL stages? (2) How do seven convolutional neural network (CNN) architectures perform across four distinct deep learning tasks? (3) What is the impact of CLAHE (Contrast Limited Adaptive Histogram Equalization) on classification performance?

Approach: We created a dataset of 14,607 annotated knee AP X-rays from three hospitals. The knee joint region was isolated using a YOLOv5 object detection model. The multi-model approach utilized three DL models: one for osteophyte detection, another for joint space narrowing analysis, and a third to combine these outputs with demographic and image data for KL classification. The single-model approach directly classified KL stages as a benchmark. Seven CNN architectures (NfNet-F0/F1, EfficientNet-B0/B3, Inception-ResNet-v2, VGG16) were trained with and without CLAHE augmentation.

Results: The single-model approach achieved an F1-score of 0.763 and accuracy of 0.767, outperforming the multi-model strategy, which scored 0.736 and 0.740. Different models performed best across tasks, underscoring the need for task-specific architecture selection. CLAHE negatively impacted most models, with only one showing a marginal improvement of 0.3%.

Conclusion: The single-model approach was more effective for KL grading, surpassing metrics in existing literature. These findings emphasize the importance of task-specific architectures and preprocessing. Future studies should explore ensemble modeling, advanced augmentations, and clinical validation to enhance applicability.

Background: Clinical data is instrumental to medical research, machine learning (ML) model development, and advancing surgical care, but access is often constrained by privacy regulations and missing data. Synthetic data offers a promising solution to preserve privacy while enabling broader data access. Recent advances in large language models (LLMs) provide an opportunity to generate synthetic data with reduced reliance on domain expertise, computational resources, and pre-training.

Objective: This study aims to assess the feasibility of generating realistic tabular clinical data with OpenAI's GPT-4o using zero-shot prompting, and evaluate the fidelity of LLM-generated data by comparing its statistical properties to the Vital Signs DataBase (VitalDB), a real-world open-source perioperative dataset.

Methods: In Phase 1, GPT-4o was prompted to generate a dataset with qualitative descriptions of 13 clinical parameters. The resultant data was assessed for general errors, plausibility of outputs, and cross-verification of related parameters. In Phase 2, GPT-4o was prompted to generate a dataset using descriptive statistics of the VitalDB dataset. Fidelity was assessed using two-sample t-tests, two-sample proportion tests, and 95% confidence interval (CI) overlap.

Results: In Phase 1, GPT-4o generated a complete and structured dataset comprising 6,166 case files. The dataset was plausible in range and correctly calculated body mass index for all case files based on respective heights and weights. Statistical comparison between the LLM-generated datasets and VitalDB revealed that Phase 2 data achieved significant fidelity. Phase 2 data demonstrated statistical similarity in 12/13 (92.31%) parameters, whereby no statistically significant differences were observed in 6/6 (100.0%) categorical/binary and 6/7 (85.71%) continuous parameters. Overlap of 95% CIs were observed in 6/7 (85.71%) continuous parameters.

Conclusion: Zero-shot prompting with GPT-4o can generate realistic tabular synthetic datasets, which can replicate key statistical properties of real-world perioperative data. This study highlights the potential of LLMs as a novel and accessible modality for synthetic data generation, which may address critical barriers in clinical data access and eliminate the need for technical expertise, extensive computational resources, and pre-training. Further research is warranted to enhance fidelity and investigate the use of LLMs to amplify and augment datasets, preserve multivariate relationships, and train robust ML models.

Anomaly detection is vital for enhancing the safety of Industrial Control Systems (ICS). However, the complicated structure of ICS creates complex temporal correlations among devices with many parameters. Current methods often ignore these correlations and poorly select parameters, missing valuable insights. Additionally, they lack interpretability, operating efficiently with limited resources, and root cause identification. This study proposes an explainable correlation-based anomaly detection method for ICS. The optimal window size of the data is determined using Long Short-Term Memory Networks-Autoencoder (LSTM-AE) and the correlation parameter set is extracted using the Pearson correlation. A Latent Correlation Matrix (LCM) is created from the correlation parameter set and a Latent Correlation Vector (LCV) is derived from LCM. Based on the LCV, the method utilizes a Multivariate Gaussian Distribution (MGD) to identify anomalies. This is achieved through an anomaly detection module that incorporates a threshold mechanism, utilizing alpha and epsilon values. The proposed method utilizes a novel set of input features extracted using the Shapley Additive explanation (SHAP) framework to train and evaluate the MGD model. The method is evaluated on the Secure Water Treatment (SWaT), Hardware-in-the-loop-based augmented ICS security (HIL-HAI), and Internet of Things Modbus dataset using precision, recall, and F-1 score metrics. Additionally, SHAP is used to gain insights into the anomalies and identify their root causes. Comparative experiments demonstrate the method's effectiveness, achieving a better 0.96% precision and 0.84% F1-score. This enhanced performance aids ICS engineers and decision-makers in identifying the root causes of anomalies. Our code is publicly available at a GitHub repository: https://github.com/Ermiyas21/Explainable-correlation-AD.

A smart grid is a structure that regulates, operates, and utilizes energy sources that are incorporated into the smart grid using smart communications techniques and computerized techniques. The running and maintenance of Smart Grids now depend on artificial intelligence methods quite extensively. Artificial intelligence is enabling more dependable, efficient, and sustainable energy systems from improving load forecasting accuracy to optimizing power distribution and guaranteeing issue identification. An intelligent smart grid will be created by substituting artificial intelligence for manual tasks and achieving high efficiency, dependability, and affordability across the energy supply chain from production to consumption. Collection of a large diversity of data is vital to make effective decisions. Artificial intelligence application operates by processing abundant data samples, advanced computing, and strong communication collaboration. The development of appropriate infrastructure resources, including big data, cloud computing, and other collaboration platforms, must be enhanced for this type of operation. In this paper, an attempt has been made to summarize the artificial intelligence techniques used in various aspects of smart grid system.

This study examines the effectiveness of the ChatMu application in improving access to information for members of Muhammadiyah, a prominent socio-religious organization. The research employs a mixed-methods approach, combining qualitative and quantitative analyses to evaluate the application's performance, usability, and user satisfaction. Findings reveal that ChatMu significantly enhances the accessibility and accuracy of Muhammadiyah-related information, highlighting its potential as an innovative tool for addressing community-specific information needs. However, several usability challenges were identified, including navigation inefficiencies and inconsistencies in content delivery. These limitations suggest the need for further refinement to optimize user experience and functionality. Despite these issues, ChatMu demonstrates strong capabilities in providing relevant and reliable information, fostering digital literacy, and supporting information dissemination within the Muhammadiyah community. The study concludes that ChatMu represents a promising application of chatbot technology in empowering communities through improved access to knowledge. Future development efforts should focus on comprehensive usability testing, maintaining information relevance, and incorporating advanced interactive features to enhance engagement. With continuous improvements, ChatMu has the potential to become an effective medium for advancing literacy and knowledge-sharing in the Muhammadiyah community.

Algorithms are involved in decisions ranging from trivial to significant, but people often express distrust toward them. Research suggests that educational efforts to explain how algorithms work may help mitigate this distrust. In a study of 1,921 participants from 20 countries, we examined differences in algorithmic trust for low-stakes and high-stakes decisions. Our results suggest that statistical literacy is negatively associated with trust in algorithms for high-stakes situations, while it is positively associated with trust in low-stakes scenarios with high algorithm familiarity. However, explainability did not appear to influence trust in algorithms. We conclude that having statistical literacy enables individuals to critically evaluate the decisions made by algorithms, data and AI, and consider them alongside other factors before making significant life decisions. This ensures that individuals are not solely relying on algorithms that may not fully capture the complexity and nuances of human behavior and decision-making. Therefore, policymakers should consider promoting statistical/AI literacy to address some of the complexities associated with trust in algorithms. This work paves the way for further research, including the triangulation of data with direct observations of user interactions with algorithms or physiological measures to assess trust more accurately.

Introduction: Mobile app development has rapidly evolved into a crucial aspect of modern technology, driving innovation across various industries and transforming user experiences globally. The dynamic nature of mobile technology requires developers to navigate a complex landscape of platforms, devices, and user requirements. Effective management and sharing of knowledge are essential to address these challenges, ensuring streamlined development processes and enhanced collaboration among stakeholders.

Methods: To this end, ontologies have emerged as powerful tools for structuring and standardizing domain-specific knowledge. This paper introduces MAD-onto, a comprehensive ontology designed specifically for the mobile app development domain. The ontology is constructed by identifying key concepts, defining classes and their hierarchies, establishing class properties, and creating instances relevant to mobile app development. To ensure robustness, the ontology is evaluated using a multi-criteria evaluation metric, focusing on consistency, completeness, conciseness, expandability, and sensitiveness. Additionally, SWRL rules are applied to validate and enforce logical constraints within the ontology.

Results: Through these rigorous evaluation methods, MAD-onto demonstrates its utility in providing a structured framework for the mobile app development lifecycle, facilitating better decision-making, collaboration, and efficiency.

Discussion: The findings highlight the significance of ontology-driven approaches in addressing the complexities of mobile app development and set a foundation for future research and advancements in this field.

The accurate classification of potato leaf diseases plays a pivotal role in ensuring the health and productivity of crops. This study presents a unified approach for addressing this challenge by leveraging the power of Explainable AI (XAI) and transfer learning within a deep Learning framework. In this research, we propose a transfer learning-based deep learning model that is tailored for potato leaf disease classification. Transfer learning enables the model to benefit from pre-trained neural network architectures and weights, enhancing its ability to learn meaningful representations from limited labeled data. Additionally, Explainable AI techniques are integrated into the model to provide interpretable insights into its decision-making process, contributing to its transparency and usability. We used a publicly available potato leaf disease dataset to train the model. The results obtained are 97% for validation accuracy and 98% for testing accuracy. This study applies gradient-weighted class activation mapping (Grad-CAM) to enhance model interpretability. This interpretability is vital for improving predictive performance, fostering trust, and ensuring seamless integration into agricultural practices.

Introduction: The integration of recent technologies in medical imaging has become a cornerstone of modern healthcare, facilitating detailed analysis of internal anatomy and pathology. Traditional methods, however, often grapple with data-sharing restrictions due to privacy concerns. Emerging techniques in artificial intelligence offer innovative solutions to overcome these constraints, with synthetic data generation enabling the creation of realistic medical imaging datasets, but the preservation of critical hidden medical biomarkers is an open question.

Methods: This study employs state-of-the-art Denoising Diffusion Probabilistic Models integrated with a Swin-transformer-based network to generate synthetic medical data. Three distinct areas of medical imaging - radiology, ophthalmology, and histopathology - are explored. The quality of synthetic images is evaluated through a classifier trained to identify the preservation of medical biomarkers.

Results: The diffusion model effectively preserves key medical features, such as lung markings and retinal abnormalities, producing synthetic images closely resembling real data. Classifier performance demonstrates the reliability of synthetic data for downstream tasks, with F1 and AUC reaching 0.8-0.99.

Discussion: This work provides valuable insights into the potential of diffusion-based models for generating realistic, biomarker-preserving synthetic images across various medical imaging modalities. These findings highlight the potential of synthetic data to address challenges such as data scarcity and privacy concerns in clinical practice, research, and education.

Understanding how language and linguistic constructions are processed in the brain is a fundamental question in cognitive computational neuroscience. In this study, we investigate the processing and representation of Argument Structure Constructions (ASCs) in the BERT language model, extending previous analyses conducted with Long Short-Term Memory (LSTM) networks. We utilized a custom GPT-4 generated dataset comprising 2000 sentences, evenly distributed among four ASC types: transitive, ditransitive, caused-motion, and resultative constructions. BERT was assessed using the various token embeddings across its 12 layers. Our analyses involved visualizing the embeddings with Multidimensional Scaling (MDS) and t-Distributed Stochastic Neighbor Embedding (t-SNE), and calculating the Generalized Discrimination Value (GDV) to quantify the degree of clustering. We also trained feedforward classifiers (probes) to predict construction categories from these embeddings. Results reveal that CLS token embeddings cluster best according to ASC types in layers 2, 3, and 4, with diminished clustering in intermediate layers and a slight increase in the final layers. Token embeddings for DET and SUBJ showed consistent intermediate-level clustering across layers, while VERB embeddings demonstrated a systematic increase in clustering from layer 1 to 12. OBJ embeddings exhibited minimal clustering initially, which increased substantially, peaking in layer 10. Probe accuracies indicated that initial embeddings contained no specific construction information, as seen in low clustering and chance-level accuracies in layer 1. From layer 2 onward, probe accuracies surpassed 90 percent, highlighting latent construction category information not evident from GDV clustering alone. Additionally, Fisher Discriminant Ratio (FDR) analysis of attention weights revealed that OBJ tokens had the highest FDR scores, indicating they play a crucial role in differentiating ASCs, followed by VERB and DET tokens. SUBJ, CLS, and SEP tokens did not show significant FDR scores. Our study underscores the complex, layered processing of linguistic constructions in BERT, revealing both similarities and differences compared to recurrent models like LSTMs. Future research will compare these computational findings with neuroimaging data during continuous speech perception to better understand the neural correlates of ASC processing. This research demonstrates the potential of both recurrent and transformer-based neural language models to mirror linguistic processing in the human brain, offering valuable insights into the computational and neural mechanisms underlying language understanding.