Search Results (72)

Background/Objectives: Accurate assessment of speech intelligibility is necessary for individuals with motor speech disorders. Transcription or scaled rating methods by naïve listeners are the most reliable tasks for these purposes; however, they are often resource-intensive and time-consuming within clinical contexts. Automatic speech recognition (ASR) systems, which transcribe speech into text, have been increasingly utilized for assessing speech intelligibility. This study investigates the feasibility of using an open-source ASR system to assess speech intelligibility in Hebrew and English speakers with Down syndrome (DS). Methods: Recordings from 65 Hebrew- and English-speaking participants were included: 33 speakers with DS and 32 typically developing (TD) peers. Speech samples (words, sentences) were transcribed using Whisper (OpenAI) and by naïve listeners. The proportion of agreement between ASR transcriptions and those of naïve listeners was compared across speaker groups (TD, DS) and languages (Hebrew, English) for word-level data. Further comparisons for Hebrew speakers were conducted across speaker groups and stimuli (words, sentences). Results: The strength of the correlation between listener and ASR transcription scores varied across languages, and was higher for English (r = 0.98) than for Hebrew (r = 0.81) for speakers with DS. A higher proportion of listener–ASR agreement was demonstrated for TD speakers, as compared to those with DS (0.94 vs. 0.74, respectively), and for English, in comparison to Hebrew speakers (0.91 for English DS speakers vs. 0.74 for Hebrew DS speakers). Listener–ASR agreement for single words was consistently higher than for sentences among Hebrew speakers. Speakers’ intelligibility influenced word-level agreement among Hebrew- but not English-speaking participants with DS. Conclusions: ASR performance for English closely approximated that of naïve listeners, suggesting potential near-future clinical applicability within single-word intelligibility assessment. In contrast, a lower proportion of agreement between human listeners and ASR for Hebrew speech indicates that broader clinical implementation may require further training of ASR models in this language. Full article

With the accelerated digital transformation of the power industry, critical infrastructures such as power grids are increasingly migrating to cloud-native architectures, leading to unprecedented growth in service scale and complexity. Traditional operation and maintenance (O&M) methods struggle to meet the demands for real-time monitoring, accuracy, and scalability in such environments. This paper proposes a novel service performance anomaly detection system based on large language models (LLMs) and multi-agent systems (MAS). By integrating the semantic understanding capabilities of LLMs with the distributed collaboration advantages of MAS, we construct a high-precision and robust anomaly detection framework. The system adopts a hierarchical architecture, where lower-layer agents are responsible for tasks such as log parsing and metric monitoring, while an upper-layer coordinating agent performs multimodal feature fusion and global anomaly decision-making. Additionally, the LLM enhances the semantic analysis and causal reasoning capabilities for logs. Experiments conducted on real-world data from the State Grid Corporation of China, covering 1289 service combinations, demonstrate that our proposed system significantly outperforms traditional methods in terms of the F1-score across four platforms, including customer services and grid resources (achieving up to a 10.3% improvement). Notably, the system excels in composite anomaly detection and root cause analysis. This study provides an industrial-grade, scalable, and interpretable solution for intelligent power grid O&M, offering a valuable reference for the practical implementation of AIOps in critical infrastructures. Evaluated on real-world data from the State Grid Corporation of China (SGCC), our system achieves a maximum F1-score of 88.78%, with a precision of 92.16% and recall of 85.63%, outperforming five baseline methods. Full article

Background/Objectives: The RTK-RAS signaling cascade is a central axis in colorectal cancer (CRC) pathogenesis, governing cellular proliferation, survival, and therapeutic resistance. Somatic alterations in key pathway genes—including KRAS, NRAS, BRAF, and EGFR—are pivotal to clinical decision-making in precision oncology. However, the integration of these genomic events with clinical and demographic data remains hindered by fragmented resources and a lack of accessible analytical frameworks. To address this challenge, we developed AI-HOPE-RTK-RAS, a domain-specialized conversational artificial intelligence (AI) system designed to enable natural language-based, integrative analysis of RTK-RAS pathway alterations in CRC. Methods: AI-HOPE-RTK-RAS employs a modular architecture combining large language models (LLMs), a natural language-to-code translation engine, and a backend analytics pipeline operating on harmonized multi-dimensional datasets from cBioPortal. Unlike general-purpose AI platforms, this system is purpose-built for real-time exploration of RTK-RAS biology within CRC cohorts. The platform supports mutation frequency profiling, odds ratio testing, survival modeling, and stratified analyses across clinical, genomic, and demographic parameters. Validation included reproduction of known mutation trends and exploratory evaluation of co-alterations, therapy response, and ancestry-specific mutation patterns. Results: AI-HOPE-RTK-RAS enabled rapid, dialogue-driven interrogation of CRC datasets, confirming established patterns and revealing novel associations with translational relevance. Among early-onset CRC (EOCRC) patients, the prevalence of RTK-RAS alterations was significantly lower compared to late-onset disease (67.97% vs. 79.9%; OR = 0.534, p = 0.014), suggesting the involvement of alternative oncogenic drivers. In KRAS-mutant patients receiving Bevacizumab, early-stage disease (Stages I–III) was associated with superior overall survival relative to Stage IV (p = 0.0004). In contrast, BRAF-mutant tumors with microsatellite-stable (MSS) status displayed poorer prognosis despite higher chemotherapy exposure (OR = 7.226, p < 0.001; p = 0.0000). Among EOCRC patients treated with FOLFOX, RTK-RAS alterations were linked to worse outcomes (p = 0.0262). The system also identified ancestry-enriched noncanonical mutations—including CBL, MAPK3, and NF1—with NF1 mutations significantly associated with improved prognosis (p = 1 × 10⁻⁵). Conclusions: AI-HOPE-RTK-RAS exemplifies a new class of conversational AI platforms tailored to precision oncology, enabling integrative, real-time analysis of clinically and biologically complex questions. Its ability to uncover both canonical and ancestry-specific patterns in RTK-RAS dysregulation—especially in EOCRC and populations with disproportionate health burdens—underscores its utility in advancing equitable, personalized cancer care. This work demonstrates the translational potential of domain-optimized AI tools to accelerate biomarker discovery, support therapeutic stratification, and democratize access to multi-omic analysis. Full article

Dementia is a neurodegenerative disorder characterized by progressive cognitive impairment that significantly affects daily living. Early detection of Alzheimer’s disease—the most common form of dementia—remains essential for prompt intervention and treatment, yet clinical diagnosis often requires extensive and resource-intensive procedures. This article explores the effectiveness of automated Natural Language Processing (NLP) methods for identifying Alzheimer’s indicators from audio transcriptions of the Cookie Theft picture description task in the PittCorpus dementia database. Five NLP approaches were compared: a classical Tf–Idf statistical representation and embeddings derived from large language models (GloVe, BERT, Gemma-2B, and Linq-Embed-Mistral), each integrated with a logistic regression classifier. Transcriptions were carefully preprocessed to preserve linguistically relevant features such as repetitions, self-corrections, and pauses. To compare the performance of the five approaches, a stratified 5-fold cross-validation was conducted; the best results were obtained with BERT embeddings (84.73% accuracy) closely followed by the simpler Tf–Idf approach (83.73% accuracy) and the state-of-the-art model Linq-Embed-Mistral (83.54% accuracy), while Gemma-2B and GloVe embeddings yielded slightly lower performances (80.91% and 78.11% accuracy, respectively). Contrary to initial expectations—that richer semantic and contextual embeddings would substantially outperform simpler frequency-based methods—the competitive accuracy of Tf–Idf suggests that the choice and frequency of the words used might be more important than semantic or contextual information in Alzheimer’s detection. This work represents an effort toward implementing user-friendly software capable of offering an initial indicator of Alzheimer’s risk, potentially reducing the need for an in-person clinical visit. Full article

Background: Electronic patient-reported outcomes (ePROs) enable real-time symptom monitoring and early intervention in oncology. Large language models (LLMs), when combined with retrieval-augmented generation (RAG), offer scalable Artificial Intelligence (AI)-driven education tailored to individual patient needs. However, few studies have examined the feasibility and clinical impact of integrating ePRO with LLM-RAG feedback during radiotherapy in high-toxicity settings such as head and neck cancer. Methods: This prospective observational study enrolled 42 patients with head and neck cancer undergoing radiotherapy from January to December 2024. Patients completed ePRO entries twice weekly using a web-based platform. Following each entry, an LLM-RAG system (Gemini 1.5-based) generated real-time educational feedback using National Comprehensive Cancer Network (NCCN) guidelines and institutional resources. Primary outcomes included percentage weight loss and treatment interruption days. Statistical analyses included t-tests, linear regression, and receiver operating characteristic (ROC) analysis. A threshold of ≥6 ePRO entries was used for subgroup analysis. Results: Patients had a mean age of 53.6 years and submitted an average of 8.0 ePRO entries. Frequent ePRO users (≥6 entries) had significantly less weight loss (4.45% vs. 7.57%, p = 0.021) and fewer treatment interruptions (0.67 vs. 2.50 days, p = 0.002). Chemotherapy, moderate-to-severe pain, and lower ePRO submission frequency were associated with greater weight loss. ePRO submission frequency was negatively correlated with both weight loss and treatment interruption days. The most commonly reported symptoms were appetite loss, fatigue, and nausea. Conclusions: Integrating LLM-RAG feedback with ePRO systems is feasible and may enhance symptom control, treatment continuity, and patient engagement in head and neck cancer radiotherapy. Further studies are warranted to validate the clinical benefits of AI-supported ePRO platforms in routine care. Full article

Energy consumption has emerged as a critical design constraint in deploying high-performance neural networks, especially on edge devices with limited power resources. In this paper, a comparative study is conducted for two prevalent deep learning paradigms—convolutional neural networks (CNNs), exemplified by ResNet18, and transformer-based large language models (LLMs), represented by GPT3-small, Llama-7B, and GPT3-175B. By analyzing how the scaling of memory energy versus computing energy affects the energy consumption of neural networks with different batch sizes (1, 4, 8, 16), it is shown that ResNet18 transitions from a memory energy-limited regime at low batch sizes to a computing energy-limited regime at higher batch sizes due to its extensive convolution operations. On the other hand, GPT-like models remain predominantly memory-bound, with large parameter tensors and frequent key–value (KV) cache lookups accounting for most of the total energy usage. Our results reveal that reducing weight-memory energy is particularly effective in transformer architectures, while improving multiply–accumulate (MAC) efficiency significantly benefits CNNs at higher workloads. We further highlight near-memory and in-memory computing approaches as promising strategies to lower data-transfer costs and enhance power efficiency in large-scale deployments. These findings offer actionable insights for architects and system designers aiming to optimize artificial intelligence (AI) performance under stringent energy budgets on battery-powered edge devices. Full article

Rendering surface materials to provide realistic tactile sensations is a key focus in haptic interaction research. However, generating texture maps and designing corresponding haptic feedback often requires expert knowledge and significant effort. To simplify the workflow, we developed a micro-collision-based tactile texture dataset for several common materials and fine-tuned the VAE model of Stable Diffusion. Our approach allows designers to generate matching visual and haptic textures from natural language prompts and enables users to receive real-time, realistic haptic feedback when interacting with virtual surfaces. We evaluated our method through a haptic design task. Professional and non-haptic designers each created one haptic design using traditional tools and another using our approach. Participants then evaluated the four resulting designs. The results showed that our method produced haptic feedback comparable to that of professionals, though slightly lower in overall and consistency scores. Importantly, professional designers using our method required less time and fewer expert resources. Non-haptic designers also achieved better outcomes with our tool. Our generative method optimizes the haptic design workflow, lowering the expertise threshold and increasing efficiency. It has the potential to support broader adoption of haptic design in interactive media and enhance multisensory experiences. Full article

This article introduces a freely available Spanish–Mixtec parallel corpus designed to foster natural language processing (NLP) development for an indigenous language that remains digitally low-resourced. The dataset, comprising 14,587 sentence pairs, covers Mixtec variants from Guerrero (Tlacoachistlahuaca, Northern Guerrero, and Xochapa) and Oaxaca (Western Coast, Southern Lowland, Santa María Yosoyúa, Central, Lower Cañada, Western Central, San Antonio Huitepec, Upper Western, and Southwestern Central). Texts are classified into four main domains as follows: education, law, health, and religion. To compile these data, we conducted a two-phase collection process as follows: first, an online search of government portals, religious organizations, and Mixtec language blogs; and second, an on-site retrieval of physical texts from the library of the Autonomous University of Querétaro. Scanning and optical character recognition were then performed to digitize physical materials, followed by manual correction to fix character misreadings and remove duplicates or irrelevant segments. We conducted a preliminary evaluation of the collected data to validate its usability in automatic translation systems. From Spanish to Mixtec, a fine-tuned GPT-4o-mini model yielded a BLEU score of 0.22 and a TER score of 122.86, while two fine-tuned open source models mBART-50 and M2M-100 yielded BLEU scores of 4.2 and 2.63 and TER scores of 98.99 and 104.87, respectively. All code demonstrating data usage, along with the final corpus itself, is publicly accessible via GitHub and Figshare. We anticipate that this resource will enable further research into machine translation, speech recognition, and other NLP applications while contributing to the broader goal of preserving and revitalizing the Mixtec language. Full article

This study presents an advanced hybrid approach for detecting near-duplicate texts in the Kazakh language, addressing the specific challenges posed by its agglutinative morphology. The proposed method combines statistical and semantic techniques, including N-gram analysis, TF-IDF, LSH, LSA, and LDA, and is benchmarked against the bert-base-multilingual-cased model. Experiments were conducted on the purpose-built Arailym-aitu/KazakhTextDuplicates corpus, which contains over 25,000 manually modified text fragments using typical techniques, such as paraphrasing, word order changes, synonym substitution, and morphological transformations. The results show that the hybrid model achieves a precision of 1.00, a recall of 0.73, and an F1-score of 0.84, significantly outperforming traditional N-gram and TF-IDF approaches and demonstrating comparable accuracy to the BERT model while requiring substantially lower computational resources. The hybrid model proved highly effective in detecting various types of near-duplicate texts, including paraphrased and structurally modified content, making it suitable for practical applications in academic integrity verification, plagiarism detection, and intelligent text analysis. Moreover, this study highlights the potential of lightweight hybrid architectures as a practical alternative to large transformer-based models, particularly for languages with limited annotated corpora and linguistic resources. It lays the foundation for future research in cross-lingual duplicate detection and deep model adaptation for the Kazakh language. Full article

Constraint satisfaction problems (CSPs) are widely used in domains such as product configuration, scheduling, and resource allocation. However, formulating constraint models remains a significant challenge that often requires specialized expertise in constraint programming (CP). This study introduces the Automatic Constraint Model Generator (ACMG), a novel framework that leverages fine-tuned large language models (LLMs) to automate the translation of natural language problem descriptions into formal CSP models. The ACMG employs a multi-step process involving semantic entity extraction, constraint model generation, and iterative validation using the MiniZinc solver. Our approach achieves state-of-the-art (SOTA) or near-SOTA results, demonstrating the viability of LLMs in simplifying the adoption of CP. Its key contributions include a high-quality dataset for fine-tuning, a modular architecture with specialized LLM components, and empirical validation which shows its promising results for complex configuration tasks. By bridging the gap between natural language and formal constraint models, the ACMG significantly lowers the barrier to CP, making it more accessible to non-experts while maintaining a high level of robustness for industrial applications. Full article

The emergence of large language models (LLMs) has revolutionized the trajectory of NLP research. Transformers, combined with attention mechanisms, have increased computational power, and massive datasets have led to the emergence of pre-trained large language models (PLLMs), which offer promising possibilities for multilingual applications in low-resource settings. However, the scarcity of annotated resources and suitably pre-trained models continues to pose a significant hurdle for the low-resource abstractive text summarization of legal texts, particularly in Urdu. This study presents a transfer learning approach using pre-trained multilingual large models (the mBART and mT5, Small, Base, and Large) to generate abstractive summaries of Urdu legal texts. A curated dataset was developed with legal experts, who produced ground-truth summaries. The models were fine-tuned on this domain-specific corpus to adapt them for low-resource legal summarization. The experimental results demonstrated that the mT5-Large, fine-tuned on Urdu legal texts, outperforms all other evaluated models across standard summarization metrics, achieving ROUGE-1 scores of 0.7889, ROUGE-2 scores of 0.5961, and ROUGE-L scores of 0.7813. This indicates its strong capacity to generate fluent, coherent, and legally accurate summaries. The mT5-Base model closely follows with ROUGE-1 = 0.7774, while the mT5-Small shows moderate performance (ROUGE-1 = 0.6406), with reduced fidelity in capturing legal structure. The mBART50 model, despite being fine-tuned on the same legal corpus, performs lower (ROUGE-1 = 0.5914), revealing its relative limitations in this domain. Notably, models trained or fine-tuned on non-legal, out-of-domain data, such as the urT5 (ROUGE-1 = 0.3912), the mT5-XLSUM (ROUGE-1 = 0.0582), and the mBART50 (XLSUM) (ROUGE-1 = 0.0545), exhibit poor generalization to legal summaries, underscoring the necessity of domain adaptation when working in low-resource legal contexts. These findings highlight the effectiveness of fine-tuning multilingual LLMs for domain-specific tasks. The gains in legal summarization demonstrate the practical value of transfer learning in low-resource settings and the broader potential of AI-driven tools for legal document processing, information retrieval, and decision support. Full article

This paper introduces a pioneering hybrid framework that integrates Q-learning/deep Q-network (DQN) with a locally deployed large language model (LLM) to enhance obstacle avoidance in embedded robotic systems. The STM32WB55RG microcontroller handles real-time decision-making using sensor data, while a Raspberry Pi 5 computer runs a quantized TinyLlama LLM to dynamically refine navigation strategies. The LLM addresses traditional Q-learning limitations, such as slow convergence and poor adaptability, by analyzing action histories and optimizing decision-making policies in complex, dynamic environments. A selective triggering mechanism ensures efficient LLM intervention, minimizing computational overhead. Experimental results demonstrate significant improvements, including up to 41% higher deadlock recovery (81% vs. 40% for Q-learning + LLM), up to 34% faster time to goal (38 s vs. 58 s for Q-learning + LLM), and up to 14% lower collision rates (11% vs. 25% for Q-learning + LLM) compared to standalone Q-learning/DQN. This novel approach presents a solution for scalable, adaptive navigation in resource-constrained embedded robotics, with potential applications in logistics and healthcare. Full article

The rapid advancement of artificial intelligence (AI) has introduced new challenges, particularly in the generation of AI-written content that closely resembles human-authored text. This poses a significant risk for misinformation, digital fraud, and academic dishonesty. While large language models (LLM) have demonstrated impressive capabilities across various languages, there remains a critical gap in evaluating and detecting AI-generated content in under-resourced languages such as Turkish. To address this, our study investigates the effectiveness of long short-term memory (LSTM) networks—a computationally efficient and interpretable architecture—for distinguishing AI-generated Turkish texts produced by ChatGPT from human-written content. LSTM was selected due to its lower hardware requirements and its proven strength in sequential text classification, especially under limited computational resources. Four experiments were conducted, varying hyperparameters such as dropout rate, number of epochs, embedding size, and patch size. The model trained over 20 epochs achieved the best results, with a classification accuracy of 97.28% and an F1 score of 0.97 for both classes. The confusion matrix confirmed high precision, with only 19 misclassified instances out of 698. These findings highlight the potential of LSTM-based approaches for AI-generated text detection in the Turkish language context. This study not only contributes a practical method for Turkish NLP applications but also underlines the necessity of tailored AI detection tools for low-resource languages. Future work will focus on expanding the dataset, incorporating other architectures, and applying the model across different domains to enhance generalizability and robustness. Full article

Music has been shown to increase arousal and attention and even facilitate processing during non-musical tasks, including those related to speech and language functions. Mind wandering has been studied in many sustained attention tasks. Here, we investigated the intersection of these two phenomena: the role of mind wandering while listening to familiar/unfamiliar musical excerpts, and its effects on concurrent linguistic processing. We hypothesized that familiar music would be less distracting than unfamiliar music, causing less mind wandering, and consequently benefit concurrent speech perception. Participants (N = 96 young adults) performed a lexical-semantic congruity task where they judged the relatedness of visually presented word pairs while listening to non-vocal classical music (familiar or unfamiliar orchestral pieces), or a non-music environmental sound clip (control) played in the background. Mind wandering episodes were probed intermittently during the task by explicitly asking listeners if their mind was wandering in that moment. The primary outcome was accuracy and reactions times measured during the lexical-semantic judgment task across the three background music conditions (familiar, unfamiliar, and control). We found that listening to familiar music, relative to unfamiliar music or environmental noise, was associated with faster lexical-semantic decisions and a lower incidence of mind wandering. Mind wandering frequency was similar when performing the task when listening to familiar music and control environmental sounds. We infer that familiar music increases task enjoyment, reduces mind wandering, and promotes more rapid lexical access during concurrent lexical processing, by modulating task-related attentional resources. The implications of using music as an aid during academic study and cognitive tasks are discussed. Full article

This paper introduces a novel framework for addressing domain adaptation challenges in large language models (LLMs), emphasising privacy-preserving synthetic data generation and efficient fine-tuning. The proposed framework employs a multi-stage approach that includes document ingestion, relevance assessment, and automated dataset creation. This process reduces the need for extensive technical expertise while safeguarding data privacy. We evaluate the framework’s performance on domain-specific tasks in fields such as biobanking and public health, demonstrating that models fine-tuned using our method achieve results comparable to larger proprietary models. Crucially, these models maintain their general instruction-following capabilities, even when adapted to specialised domains, as shown through experiments with 7B and 8B parameter LLMs. Key components of the framework include continuous pre-training, supervised fine-tuning (SFT), and reinforcement learning methods such as direct preference optimisation (DPO), which together provide a flexible and configurable solution for deploying LLMs. The framework supports both local models and API-based solutions, making it scalable and accessible. By enabling privacy-preserving, domain-specific adaptation without requiring extensive expertise, this framework represents a significant step forward in the deployment of LLMs for specialised applications. The framework significantly lowers the barrier to domain adaptation for small- and medium-sized enterprises (SMEs), enabling them to utilise the power of LLMs without requiring extensive resources or technical expertise. Full article