Large language models (LLMs) are increasingly explored in radiology, yet concerns persist regarding hallucination and lack of factual grounding. Retrieval-augmented generation (RAG) seeks to address these limitations by coupling generative models with external knowledge retrieval. We conducted a scoping review to characterize how RAG systems have been applied in radiology and medical imaging. A systematic search of PubMed, Embase, Scopus, IEEE Xplore, and arXiv identified 45 studies implementing RAG-based approaches in radiology-related tasks. In terms of clinical tasks, RAG was most commonly applied to radiology report generation and question answering. Dense retrieval strategies predominated, while sparse, hybrid and proprietary retrieval approaches were less frequent. External knowledge sources most frequently comprised biomedical literature databases and clinical guidelines. Applications were heavily skewed toward chest radiography and X-ray-based tasks, with relatively few studies addressing CT, MRI, PET, ultrasound, or under-represented subspecialties such as pediatric radiology and neuroradiology. Most comparative studies reported task-specific performance gains with RAG over non-retrieval baselines, and a small subset reported performance comparable to trained radiologists or state-of-the-art models. However, hallucinations and errors persisted, and heterogeneity across studies limited the generalizability of these findings. Evaluation practices largely relied on automated accuracy or text-overlap metrics, with limited use of standardized expert evaluation and minimal assessment of safety, bias, computational efficiency, or clinical utility. Overall, while RAG shows promise for improving factual grounding in radiology AI, current evaluation paradigms likely overestimate real-world clinical readiness. Future work should prioritize retrieval quality, clinically grounded evaluation, safety-critical error analysis, bias assessment, and deployment-relevant efficiency metrics to enable responsible clinical translation.
Continuous monitoring of chronic diseases with mobile health tools is rapidly advancing to improve patient care. Wearable sensors placed on or near the skin remain the dominant paradigm for collecting physiological data noninvasively during activities of daily living. However, wearable sensors have their limitations. Epidermal sensors are fundamentally limited by the information barrier that the skin presents. Furthermore, environmental factors impact the stability and longevity of wearables. An alternative technology is the expanding class of minimally invasive subdermal sensors, also called insertables. Key applications for subdermal sensors are reviewed, including three current clinical devices: insertable cardiac monitors, insertable continuous glucose monitors, and sub-scalp epilepsy monitors. Next, common challenges impacting performance are discussed, including device migration, device encapsulation, and misalignment with wirelessly linked wearables. Finally, recent technological advancements to mitigate these challenges are covered. With state-of-the-art approaches for achieving biocompatibility, remote powering, and wireless data transmission, the next generation of subdermal biosensors can be developed for unmet needs in chronic disease monitoring.
Soft skins with reversible thickness morphing represent a distinct and underexplored class of adaptive material interfaces. Unlike conventional soft actuators that achieve motion through bending, elongation, or twisting, these systems enable out-of-plane deformation, producing localized protrusion, retraction, and programmable contact mechanics without rigid support structures. This review reframes thickness modulation not merely as an actuation outcome, but as a material-architecture strategy that couples energy transduction, geometry, and compliance to enable new modes of haptic interaction, morphological adaptation, and operation in confined or unstructured environments. We present a comprehensive synthesis of thickness-morphing soft skins, covering actuation stimuli, material platforms, structural architectures, fabrication strategies, modeling frameworks, and system-level integration. Particular emphasis is placed on hierarchical elastomer composites, origami- and kirigami-inspired designs, electrohydraulic and multimodal hybrid systems, and emerging data-driven control approaches that expand the functional design space. Despite rapid progress, key challenges remain in durability under cyclic loading, energy efficiency and autonomy, scalable manufacturing, and integration of sensing, actuation, and computation. Addressing these challenges will enable self-powered, fault-tolerant, and computationally intelligent soft skins capable of embodied perception and safe autonomous operation, positioning thickness morphing as a foundational design axis for next-generation haptics and soft robotic systems.
Artificial Intelligence (AI) is rapidly transitioning from experimental research to daily medical practice, yet the medical community's understanding of these tools remains largely confined to visible 'front-end' applications with which the clinician can directly interact (such as decision support systems and conversational agents). This perspective overlooks the proliferation of 'back-end' AI, the algorithms that are embedded within devices, systems, and hospital infrastructure that silently reconstruct data while remaining invisible to clinicians. This paper presents a clinician-oriented framework that distinguishes between these two categories and proposes a clinically oriented guide for their evaluation. We performed a narrative review of AI applications in minimally invasive therapy and medical imaging to categorize systems into 'front-end' (interactive/visible) and 'back-end' (embedded/invisible) modalities. We synthesized evaluation metrics from computer science and engineering literature, selecting those relevant for clinical safety and decision-making to create a practical literacy guide. Front-end and back-end systems require distinct validation strategies to ensure safety. while front-end evaluation must prioritize decision quality, spatial precision, and human-computer interaction to mitigate risks like automation bias, back-end evaluation requires rigorous technical benchmarking of signal fidelity and temporal latency to ensure that algorithmic reconstruction does not distort clinical reality. To facilitate this, we developed a structured inquiry framework to guide clinicians in auditing these systems for data provenance, transparency, and failure modes. Crucially, we emphasize that mathematical optimization does not guarantee clinical efficacy; technical metrics must always be paired with specific clinical contexts to ensure they align with patient-centered outcomes. Clinical safety in the AI era demands 'algorithmic literacy'. By applying this front-end/back-end framework and understanding key technical metrics, medical professionals can better identify failure modes, ensure data integrity, and maintain clear lines of clinical accountability, shifting from passive consumers to active evaluators of medical technology.
While the graphical user interface (GUI) is fundamental to computer-aided diagnosis (CAD) systems, a significant void exists in the technical literature regarding its design and integration into clinical workflows. Bridging this gap is essential, as a well-designed interface is the key driver of usability, interpretability, and the ultimate adoption of medical imaging tools. This scoping review aims to identify and analyze the primary approaches and emerging trends in GUI design for CAD systems developed for medical image analysis, with a specific focus on classification and segmentation tasks. We analyze not only the interaction patterns but also the specific clinical applications, datasets, AI algorithms, and evaluation protocols reported in the literature. A literature search was conducted across five major digital databases (ACM Digital Library, IEEE Xplore, PubMed, ScienceDirect, and SpringerLink) for scientific publications between 2020 and 2024. The search string ("medical image" AND ("computer-aided diagnosis" OR "CADx" OR "CAD") AND "user interface") was used. The search terms were strategically selected to specifically target visual diagnostic AI systems that integrate a clinician-facing interface. Initial screenings, followed by the application of predefined inclusion and exclusion criteria, were performed. Data was charted using eight research questions. From an initial pool of 1147 articles identified across two search phases, a total of 46 studies met the final inclusion criteria and composed the review set. The data extraction process revealed a growing trend towards the integration of interactive machine learning features, visualization of model uncertainty, and tools for explainable AI (XAI) directly within the user interface, moving beyond simple image display and result presentation. The design of GUIs for modern CAD systems is evolving from static displays to interactive and collaborative platforms. Many of them not only present the AI's prediction but also provide clinicians with tools to understand, question, and refine the automated analysis. A clear gap remains in the standardization of usability and human-computer interaction metrics for evaluating these systems, suggesting a critical direction for future research.
Robotic technologies are increasingly used in dentistry to improve procedural precision and standardization. However, clinical implementation remains heterogeneous due to differences in autonomy, technological maturity, and available clinical evidence. This review evaluates current applications using a dual-perspective framework. A narrative review of the literature was conducted using PubMed/MEDLINE, Scopus, and Web of Science. English-language publications from 2004 onwards were considered and selected according to their relevance to robotic system design, clinical applications, autonomy levels, technical performance, and translational challenges in dentistry. Robotic applications have expanded across surgical, therapeutic, educational, and preventive domains. Most systems operate at assistive or semi-autonomous levels (Level 1-2), with the strongest clinical evidence observed in implant dentistry, where improved placement accuracy has been reported. Clinical adoption remains limited by high costs, technical complexity, and insufficient long-term outcome data. Dental robotics has evolved into a clinically applicable component of digital dentistry, but routine clinical implementation remains heterogeneous across dental specialties. The proposed dual-perspective framework facilitates evaluation of robotic technologies and their clinical readiness, while further clinical validation is needed to support routine implementation.
In many medical specialties, robot-assisted surgery has become a crucial option for surgical interventions. However, its adoption in vascular surgery remains limited, with this discipline lagging behind other fields in both research and application of this technology. The rapid advancement of robot-assisted surgery presents both an opportunity and a challenge for the development of vascular surgery. Bibliometrics can systematically and quantitatively summarize research achievements, key research directions, and emerging trends in this field, thereby guiding future research efforts on the application of robot-assisted surgery in vascular surgery. All eligible literature in this research was retrieved from the Web of Science Core Collection, covering publications released from 1997 to 2025 that focus on the clinical applications of robot-assisted surgery within vascular surgery. Two bibliometric software programs, VOSviewer 1.6.20 and CiteSpace 6.4.R2, were adopted to implement multiple analytical procedures, including author collaboration analysis, literature co-citation analysis and keyword co-occurrence analysis. Meanwhile, the citation burst detection function embedded in CiteSpace was applied to pinpoint prevailing research hotspots and emerging frontiers in this discipline. This study included 375 eligible publications from 1,777 researchers across 503 institutions in 39 countries. Annual publication output showed a steady increase, with two notable peaks in 2022 and 2025 after a marked rise post-2020. The United States led globally with 130 papers (34.6%) and 3,687 citations, followed by China, Japan, Italy, and the UK. The Beijing Institute of Technology was the most prolific institution. Key scholars including Shuxiang Guo, Norihiko Ishikawa, and Go Watanabe shaped the field's foundation, while the International Journal of Medical Robotics and Computer Assisted Surgery and The Annals of Thoracic Surgery were the primary publishing venues. Keyword clustering identified five major research themes: robotic surgery, angioplasty, force sensing, abdominal aortic aneurysm, and percutaneous coronary intervention. Temporal keyword analysis revealed a clear shift from traditional open cardiovascular procedures (e.g., coronary artery bypass grafting, laparoscopic aortic bypass) toward robotic-assisted endovascular and extravascular interventions, such as percutaneous coronary intervention and integrated robotic revascularization. Over the past twenty-nine years, China's research institutions and scholars have achieved certain accomplishments in the application research of robot-assisted surgical techniques in the field of vascular surgery. Vascular surgery has evolved from traditional open cardiovascular procedures to novel robot-assisted surgical methods suitable for both intravascular and extravascular operations.
The Internet of Things (IoT) and its applications are increasing rapidly over the years. Due to the wide variety of IoT applications, cyber attackers are exploring strong attacking methods and patterns to damage the IoT networks in real-time applications even if the IoT network is secure. To protect the IoT networks, it is essential to design and develop a real-time intrusion detection system that can detect the attacking patterns and methods and prevent them immediately. To achieve this goal, we have proposed an intrusion detection system using multiscale attention 1D convolutional neural networks for efficient detection of all major attacks. Our proposed mechanism integrates multi-scale convolutional kernels with a dual attention mechanism for computationally efficient intrusion detection. This mechanism has extracted spatial features to discriminate against the normal and malicious IoT traffic patterns. The experiment evaluation of the proposed work has tested two datasets, UNSW-NB15 and UM-NIDS 24, to evaluate its inference efficiency and intrusion detection capability. The proposed IDS has demonstrated the best performance in comparison to state-of-the-art models and achieved an accuracy of 91.03% on the UM-NIDS and an accuracy of 99.37% on the UNSW-NB15. Based on the experimental results and analysis, we can conclude that the proposed IDS with MA-1D-CNN is a lightweight, feature-efficient, and high-precision model for the real-time attack detection in large-scale IoT networks.
Intermittent fasting (IF) has emerged as a promising dietary approach with prospective advantages for clinical as well as non-clinical applications. Research indicates that IF enhances insulin sensitivity, facilitates weight reduction and stimulates cellular repair pathways, including autophagy. Physiological adaptations to fasting are reflected in favorable alterations in biomarkers and metabolic processes. This review examines the current evidence on IF by analyzing studies retrieved through schematic searches of the MEDLINE via PubMed database, Embase and ScienceDirect using specific keyword combinations. It focuses on commonly practiced regimens- Time-restricted eating (TRE) (16/8 method), Alternate-day fasting (ADF), the 5:2 intermittent energy-restriction diet and One meal a day (OMAD) approaches and explores their effects on cardiovascular function, metabolic regulation, cognitive performance and longevity. Various IF regimens including the TRE (16/8 method), ADF, the 5:2 diet and OMAD approaches are discussed in relation to their effects on cardiovascular health, cognitive function, metabolic regulation, aging and longevity. While most finding highlight significant health benefits, inconsistencies and methodological limitations are also reported. Mechanistically, IF orchestrates a coordinated metabolic response through modulation of key nutrient sensing pathway such as AMP activated protein kinase (AMPK), mechanistic target of rapamycin (mTOR) and unc-51-like kinase 1 (ULK1). These cascades interact with Sirtuins (SIRT1/3), peroxisome proliferator activated receptor gamma coactivator-1α (PGC-1α) and the transcription factor EB (TFEB) to regulate autophagy, mitochondrial biogenesis, oxidative stress defense and cellular repair. Clinically, these molecular events underpin improvements in glycaemic control, lipid metabolism and inflammatory balance, supporting the therapeutic potential of IF for cardiometabolic disorders, neuroprotection and healthy aging.
Motor imagery-based brain-computer interface (MI-BCI) applications in stroke rehabilitation aim to match brain activity with real-time feedback, thereby establishing closed-loop neural pathways and providing a basis for evaluating patients' neuroplasticity changes. Thus, constructing EEG datasets under MI paradigms is crucial for optimizing MI-BCI systems and understanding the neural rehabilitation process. However, the current limitations of single MI paradigms and the lack of relevant EEG datasets may restrict the accurate interpretation and effective application in stroke rehabilitation. This study collected EEG data from 24 stroke patients during MI tasks, including a novel "sixth finger" MI and an affected-hand MI paradigm. The dataset comprehensively covers the complete longitudinal stages of stroke rehabilitation: pre-training, post-training, and follow-up periods. The data materials include: (1) raw EEG data, (2) preprocessed data, and (3) patient clinical information. Preliminary analysis using classical machine learning algorithms (CSP + SVM and CSP + LDA) demonstrated an average classification accuracy between the two MI paradigms maintained at approximately 85%~86%. We anticipate that this dataset will facilitate research on MI-BCI paradigms and neuroplasticity for stroke, and contribute to the development of high-efficiency MI-BCI systems in the field of stroke rehabilitation.
Globally, 1.3 billion tons of food is lost or wasted each year, negatively impacting food security, the economy, and the climate. Fresh fruits and vegetables (FFVs), with their short shelf life and temperature sensitivity, are the most affected. This study systematically evaluates the integration of Machine Learning (ML), Adaptive Learning (AL), the Internet of Things (IoT), and Fog computing for temperature-break detection and prediction in FFVs supply chains. It critically evaluates their individual and combined capabilities, identifying compounding barriers that prevent genuine real-time integration of these technologies, while assessing their performance and operational readiness for real-time cold chain monitoring. Additionally, the role of fog computing in enabling efficient ML/AL deployment at the network edge is investigated for real-time applications in dynamic environments, with an implementation framework provided. Based on the PRISMA framework, searches of Scopus, Web of Science, IEEE Xplore, ACM Digital, supplemented by citation and reference chasing, produced 830 pre?deduplication records, identifying 14 relevant studies. From the 12 analysed unique-dataset studies, 7 (58.3%) collected data using Basic Sensors, 3 with WSN (25%), and 2 with IoT (16.7%). Of the 14 ML studies, 5 (35.7%) detected temperature breaks, 5 (35.7%) predicted FFVs' temperature values, 3 (21.4%) predicted internal temperature (IT) values of a cold room or container, and 1 predicted IT values and time-to-temperature breaks. None of the studies predicted temperature breaks (event occurrence) or even their causes, while a few detected these breaks and their predefined causes. Four (28.6%) studies used IoT data, but none enabled live ML inference. None of the reviewed studies includes Fog or AL. An integrated IoT-Fog-AL framework is thus proposed to address these gaps. These areas require focus to proactively reduce temperature breaks, thereby minimising food wastage and its associated effects, while also enhancing supply chain resilience and food security.
The way our eyes move while reading provides valuable insights into both the reader's cognitive processes and the properties of the text. In particular, eye-tracking-while-reading data has been shown to be highly beneficial in various technological applications, such as enhancing and interpreting neural language models and inferring a reader's characteristics. However, these applications often rely on large-scale, data-driven models, which demand extensive eye-tracking datasets that are challenging to obtain due to the resource-intensive nature of data collection. To address the challenge of data scarcity, we develop Eyettention II, an end-to-end trained deep-learning model capable of generating realistic scanpaths consisting of a complete set of fixation attributes in chronological order, including fixation location, within-word landing position, and fixation duration. Our model is lightweight, efficiently trainable on limited GPU resources, and closely aligned with cognitive theories. We demonstrate that Eyettention II surpasses state-of-the-art models in scanpath prediction and mirrors human-like gaze behavior by capturing key psycholinguistic phenomena. With its robust performance, Eyettention II holds the potential to drive advancements in natural language processing, facilitate piloting the materials of psycholinguistic experiments, and uncover new insights beyond what is explicitly encoded in theoretical cognitive models. Our code and accompanying tutorial are publicly available at https://osf.io/3ucq7/ .
Achieving a sustainable energy system for space missions remains challenging due to the continued reliance on Earth-supplied materials. This underscores the importance of in situ resource utilization (ISRU) strategies that convert planetary resources into functional electronic components. In this work, we harness the dielectric characteristics of Martian regolith (MR) simulant to create an MR/polydimethylsiloxane (PDMS) composite film with enhanced triboelectric properties. Structural and morphological analyses of the MR reveal multiple oxide-rich phases, which improve both the dielectric properties and the surface microstructure of the MR/PDMS composite film. The resultant MR/PDMS composite film-based triboelectric nanogenerator (TENG) delivers an approximately two-fold increase in open-circuit voltage compared to the pristine PDMS-based TENG. The real-world use of the MR/PDMS TENG is further demonstrated by proof-of-concept applications: a glove-mounted tactile surface sensor with wireless signal transmission and a wearable triboelectric keypad. This work not only showcases advances in MR-based TENG performance but also marks the first demonstration of triboelectric applications using MR simulants as functional triboelectric material. Additionally, we have demonstrated foundational work toward ISRU-oriented tactile interfaces incorporating MR-simulant-derived functional materials for future controlled habitats and robotic platforms relevant to future space exploration.
Adverse drug events (ADEs) remain a major cause of preventable healthcare complications due to incorrect pill identification, dosage errors, and confusion between visually identical pills, particularly among older adults, visually impaired individuals, and people with limited health literacy. Recent advances in artificial intelligence and computer vision have enabled automated pill recognition systems. However, many existing methods address detection, classification, and imprint recognition as isolated tasks without providing a lightweight unified framework suitable for real-time healthcare deployment. This study presents a lightweight hybrid framework for multi-pill and multi-attribute recognition to support pharmacovigilance and artificial intelligence-driven clinical decision support. The proposed framework integrates pill detection using YOLOv26n, MobileNetV4-ConvSmall-based classification, EasyOCR-based imprint recognition, and metadata retrieval into a unified prediction pipeline. A customized benchmark dataset containing 400 pill classes, each with 50 images, for a total of 20,000 images, was used for classification. Each image has two pills, comprising 40,000 manually labeled pill instances used for detection. Comparative experiments, functional comparisons with existing pill recognition systems, and cross-device qualitative evaluations were conducted to assess framework performance, robustness, and deployment feasibility. YOLOv26n achieved a precision of 0.963, a recall of 0.982, mAP@50 of 0.989, and mAP@50-95 of 0.981, while MobileNetV4-ConvSmall achieved a class accuracy of 99.60% with higher performance in pill shape and color accuracy. Comparative optical character recognition (OCR) analysis shows that EasyOCR achieved more reliable imprint recognition performance than TesseractOCR and TrOCR. Comparative framework evaluation demonstrated that the YOLOv26n with MobileNetV4-ConvSmall combination achieved the high overall recognition performance, obtaining a Top-1 accuracy of 89.83% and a Top-5 accuracy of 99.49%. Cross-device qualitative evaluation using smartphone cameras and direct laptop uploads demonstrated robust performance under varying image noise conditions. The proposed framework demonstrates the potential of lightweight deep learning architectures for scalable medication identification and AI-assisted pharmacovigilance applications. By integrating detection, classification, imprint recognition, and metadata retrieval into a unified framework, the proposed system has the potential to support medication identification and improve assistive healthcare workflows. The final clinical judgment and medication verification should remain under the supervision of a healthcare professional.
Nanoparticles are important in materials science and engineering, with applications in biomedicine and electronics. Understanding their physical and structural properties relies on electron microscopy imaging techniques. Traditionally, these images have been analyzed manually, which is costly, requires specialized expertise, and limits throughput. Recently, artificial intelligence, particularly deep learning, has emerged as a powerful alternative for image analysis, enabling automated detection and segmentation. However, most existing models are designed for general-purpose tasks, and few studies focus on nanoparticle segmentation in electron microscopy images. Moreover, implementing such models often requires significant computational resources and large annotated datasets, which may not be accessible in all research settings. This study evaluates deep-learning segmentation models for identifying facets and volumes in simulated images of gold nanoparticles. Using transfer learning, we assess YOLOv11n-seg as a lightweight instance-segmentation model for facet and volume segmentation in simulated CTEM images of gold decahedral and icosahedral nanoparticles. Its use was motivated by its compact architecture within the YOLO segmentation family and its potential for efficient inference relative to larger YOLO segmentation variants. Because only simulated CTEM data are used, the reported performance should be interpreted as simulation-domain performance; validation on experimental micrographs remains necessary before practical deployment.
Digital applications based on health care conversational agents (HCAs) are increasingly being developed to support health care provision. The usability and user experience of these solutions are critical determinants of their acceptability and, consequently, their impact on health-related outcomes. This systematic review aims to synthesize current evidence on the use of valid and reliable subjective instruments for assessing the usability and user experience of HCAs and to examine whether assessment outcomes vary according to their technical characteristics. A systematic search was conducted in PubMed, Web of Science, and Scopus from inception to February 2026. Studies were included if they used subjective instruments to evaluate the usability or user experience of HCAs. A total of 127 studies met the inclusion criteria. The studies examined 3 categories of HCAs-text-based, voice-based, and embodied-applied to patient care, health education and prevention, health data collection, and support for daily activities among older adults. The System Usability Scale (SUS) was the most frequently used assessment instrument. Comparative analysis of SUS scores indicated higher usability ratings for text-based HCAs relative to voice-based and embodied systems. However, SUS and other subjective instruments used in the included studies may not fully capture key dimensions of usability and user experience of HCAs. Additionally, substantial heterogeneity was observed in assessment methodologies across studies. Comparative analysis suggested that text-based HCAs were associated with significantly higher SUS scores than voice-based and embodied HCAs. However, this finding should be interpreted with caution given the substantial heterogeneity across the included studies in health care application domains, study designs, evaluation contexts, participant populations, and HCAs' implementation and use characteristics, as well as the limitations of the SUS in evaluating the usability of modern HCAs. The variability in assessment approaches underscores the need for standardized protocols and the development of more context-specific evaluation frameworks to enhance methodological consistency and comparability across studies.
Ultraviolet (UV) imagers are important for a variety of applications, such as quality inspection in the semiconductor industry, forensics and food quality inspection, but are often costly because they require dedicated semiconductor process flows. Here, an imaging chip is introduced that has been fabricated using standard 40 nm complementary metal-oxide-semiconductor (CMOS) technology. Instead of using a conventional charge-based photodetection principle, the imager uses a capacitive operation principle where UV-light causes capacitance changes via the photodielectric effect in a functionalization layer, which are measured by the underlying CMOS circuitry. This spin-coated or inkjet-printed functionalization layer consists of solution-processed, wide-bandgap, semiconducting metal-oxide nanoparticles, such as ZnO, SnO2 and Ga2O3. Owing to their bandgap-dependent optical absorption, these materials exhibit distinct capacitive responses across UV-A, UV-B, and UV-C spectral regions, thereby enabling band-selective detection and multispectral UV imaging. The sensors exhibit low noise-equivalent powers (17-138 fW Hz-1/2) across the UV bands. Unlike conventional silicon CMOS imagers, the present capacitive-CMOS platform is inherently visible-blind, providing selective UV detection. This work positions late-functionalized capacitive-CMOS arrays as a route toward reducing the fabrication complexity of UV imagers, which can lead to their more widespread implementation in consumer and low-volume application-specific products.
The integration of CRISPR Cas genome editing with artificial intelligence (AI) offers significant potential for crop biotechnology by supporting more precise and adaptive strategies for trait improvement under complex agricultural and regulatory conditions. However, the global governance of gene edited crops remains highly heterogeneous, creating major challenges for the development of frameworks that can jointly support optimization, uncertainty management, and regulatory alignment. Conventional approaches often lack the ability to account for evolving regulatory requirements and multi source uncertainties in a unified manner. In this paper, we introduce the Adaptive Regulatory Optimizer (ARO), an AI enhanced framework designed to support CRISPR Cas genome editing in crop biotechnology under biologically, regulatorily, and contextually constrained conditions. The ARO consists of three interconnected modules: the Manifold Constrained Gene Editor, the Agent Driven Regulatory Planner, and the Uncertainty Propagation Filter. Together, these modules embed editing decisions within biologically feasible manifolds, incorporate jurisdiction aware regulatory planning, and model interacting uncertainties associated with gene editing and deployment contexts. The The framework combines constrained optimization refinement, probabilistic uncertainty modeling, and adaptive regulatory planning to provide a structured basis for compliance aware and context sensitive decision support. Experimental results on the evaluated datasets indicate that the ARO achieves improved performance on the selected metrics relative to the compared methods, while its architecture is explicitly designed to integrate regulatory constraints into the optimization process. These findings suggest that the proposed framework provides a promising foundation for supporting more transparent, adaptive, and analytically grounded decision making in CRISPR Cas applications for crop biotechnology.
Given the increasing demand for sustainable energy and self-powered devices, energy-harvesting technologies, such as triboelectric nanogenerators (TENGs), are drawing attention. To address the short charge retention in conventional polymer materials, 2D materials with high surface areas and intrinsic charge-trapping capabilities, such as MoS2, are being utilized as friction layers in TENGs. However, their power generation is too low for practical applications owing to their atomically thin nature, limiting their use as fillers in polymer-based systems. We report a 2D-material-based high-power TENG using 3D hierarchical MoS2 (3DH-MoS2) as a primary friction material. The 3DH-MoS2 is synthesized via low-temperature metal-organic chemical vapor deposition, enabling the direct growth of a uniform, large-area, 3D-nanostructured friction layer on a polymer substrate without additional processes. This 3D nanostructure increases the amount of charge-trapping sites and significantly enhances the durability of the device. The 4 × 4 cm2 3DH-MoS2-based TENG produces a maximum output voltage of 320.1 V and a power density of 0.841 mW cm-2, proving it can effectively power light-emitting diodes and a calculator, maintaining its performance over 10 000 cycles. In addition, the device can generate electricity from gas and water flow and human motion, highlighting the potential and versatility of 2D-material-based energy-harvesting systems.
BackgroundBurst deep brain stimulation (DBS) is a promising alternative to conventional deep brain stimulation (cDBS), incorporating paradigms such as theta burst stimulation (TBS), burst cycling (BC), and coordinated reset (CR). Findings from transcranial magnetic stimulation, spinal cord stimulation and focused ultrasound neuromodulation inspire its potential application in Parkinson's disease (PD).ObjectiveThis review evaluates the evidence on the efficacy and safety of burst DBS in PD.MethodsA systematic literature search was conducted in PubMed, Embase, and Web of Science. Inclusion criteria encompassed peer-reviewed primary preclinical or clinical studies in English, reporting motor outcomes and burst stimulation parameters. Methodological quality was assessed using SYRCLE and a modified Newcastle-Ottawa Scale. Data extraction was performed systematically, and findings were synthesized narratively.Results19 studies met inclusion criteria: nine preclinical and ten clinical studies. Eight focused on BC (three including safety), four examined TBS (two including safety), and seven addressed CR (one including safety). Burst DBS showed mostly similar acute efficacy to cDBS, with distinct post-stimulation effects. No major adverse events were reported. However, stimulation settings varied widely, with no consensus on optimal parameters.ConclusionEfficacy and safety of burst DBS appear to be comparable to cDBS, with potential benefits such as post-stimulation effects. However, the risk of bias, variability in stimulation settings, inconsistent terminology, and other methodological considerations limit the interpretation of previous research. Further randomized studies are crucial to better refine the stimulation parameters and establish standardized clinical protocols that can be used to compare burst DBS with cDBS. Burst deep brain stimulation for Parkinson’s disease: What current research shows about its benefits and safetyPlain language summaryParkinson’s disease affects many people worldwide and causes movement problems such as slowness, stiffness, and tremor. Deep brain stimulation (DBS) is an established treatment in which electrodes deliver electrical stimulation to specific brain regions. However, not all patients benefit equally, prompting interest in alternative stimulation approaches. Burst DBS delivers stimulation in short, patterned bursts rather than as a continuous stream and may interact with brain activity in a different, potentially more natural way. This study aimed to evaluate whether burst DBS is as effective and safe as conventional DBS for treating movement symptoms in Parkinson’s disease. Researchers also explored whether burst DBS could provide additional benefits, such as effects that persist after stimulation is switched off. A systematic review was conducted, including animal and human studies that investigated burst DBS and reported motor outcomes. The researchers searched multiple scientific databases and assessed the quality and reliability of the available evidence. Nineteen studies were included. Overall, burst DBS improved motor symptoms to a similar extent as conventional DBS, and some studies reported benefits that continued after stimulation ceased. No major safety concerns were identified. However, considerable variation in stimulation settings limited comparisons across studies. While burst DBS appears promising, further high-quality studies are needed before it can be confidently applied in clinical practice.