Human Intervention Research Articles (Page 1)

Mycotoxin exposure contributes to adverse human health outcomes, however, data regarding validated human biomarkers of exposure are lacking. This study presents an integrated framework for the biomarker discovery and toxicokinetic characterization of mycotoxin in humans. The aim of the study is to identify new biomarkers, determine their toxicokinetic (TK) properties, and build an integrated data analysis workflow using machine learning (ML), whilst focusing on non- and minimally-invasive sampling strategies. Following sample collection and chemical analysis, obtained datasets are used for the computation of ML models. Probability-based techniques are employed to calculate specific boundaries in the multidimensional space and, in parallel, ML classification methodologies are evaluated to scrutinize controls from intervened volunteers. Furthermore, multivariate regression models are computed to study the correlation of potential biomarkers with mycotoxin dosages. Once biomarkers have been identified, data are fit using Bayesian methods to a population-TK model to estimate key parameters related to absorption, distribution, metabolism, and excretion. This standardized framework allows the scientific community to identify and validate new mycotoxin biomarkers and related ADME-properties in both a precise and accurate manner. Although we developed the proposed trial for various different mycotoxins, due to ethical considerations, focus was set towards IARC group III-classified mycotoxins.

Abstract Rapid and accurate road damage detection is a critical task for road maintenance departments. Existing detection methods primarily adopt semi-automated approaches that rely on human intervention, leaving room for improvement in both efficiency and accuracy. This study utilizes vehicle-mounted cameras for pavement data collection and achieves automated detection of road damage through an improved YOLOv11s model. Key model improvements include: Firstly, replacing the SPPF module with SPPF_LSKA to enhance recognition capability in complex scenarios; Secondly, designing the M-Head-T4 structure to significantly boost detection accuracy; Thirdly, adopting the Focal-DIoU loss function to optimize the training process; Finally, conducting comparative experiments across multiple public road damage datasets. Experimental results demonstrate that the improved model achieves a 5.2% improvement in mAP@0.5 and a 4.7% increase in F1-score compared to the original YOLOv11s baseline on the SVRDD dataset.

This study investigates the impact of Natural Language Processing (NLP) on enhancing customer experience within the Zimbabwean banking industry. Employing a mixed-methods approach with a survey of 385 customers and interviews with bank officials, the research assesses the adoption, effectiveness and challenges of NLP technologies like chatbots and virtual assistants. Quantitative results indicate that NLP users reported significantly higher customer satisfaction scores and a 30% improvement in response times, compared to non-users, demonstrating NLP's efficacy in streamlining service delivery. However, first-contact resolution rates remained similar between groups, suggesting complex issues still require human intervention. Qualitative findings reveal significant implementation barriers, including high costs, a shortage of skilled personnel and integration challenges with legacy systems. Despite these hurdles, the study concludes that NLP holds substantial potential for improving operational efficiency and customer engagement in Zimbabwe. It recommends strategic investments in localized NLP solutions, capacity building and supportive policy frameworks to overcome existing challenges and fully leverage NLP for financial inclusion and competitive advantage in an evolving digital landscape.

The exponential growth of large-scale telescope arrays has boosted time-domain astronomy development but introduced operational bottlenecks, including labor-intensive observation planning, data processing, and real-time decision-making. Here we present the StarWhisper Telescope system, an AI agent framework automating end-to-end astronomical observations for surveys like the Nearby Galaxy Supernovae Survey. By integrating large language models with specialized function calls and modular workflows, StarWhisper Telescope autonomously generates site-specific observation lists, executes real-time image analysis via pipelines, and dynamically triggers follow-up proposals upon transient detection. The system reduces human intervention through automated observation planning, telescope controlling and data processing, while enabling seamless collaboration between amateur and professional astronomers. Deployed across Nearby Galaxy Supernovae Survey's network of 10 amateur telescopes, StarWhisper Telescope has detected transients with promising response times relative to existing surveys. Furthermore, StarWhisper Telescope's scalable agent architecture provides a blueprint for future facilities like the Global Open Transient Telescope Array, where AI-driven autonomy will be critical for managing 60 telescopes.

PathViT: Automated disease classification from skeletal muscle histopathology.

Agrotourism is a form of tourism that utilizes agricultural land and supporting facilities to attract visitors. One of the emerging agrotourism destinations in Rasau Jaya Tiga is Taman Inspirasi Strawberry (Strawberry Inspiration Park), which serves as a local agriculture-based tourist attraction. With technological advancements in the agricultural sector, innovations such as the Internet of Things (IoT) a concept that connects physical devices to the internet to automatically collect and exchange data without human intervention have enabled the implementation of automated plant irrigation systems. In this Community Service Program, an IoT-based automatic irrigation system was designed and implemented for strawberry cultivation. The system employs soil moisture and temperature sensors to regulate watering automatically and can be monitored via mobile devices. In addition to system installation, outreach and training sessions were conducted for park managers and participants to ensure proper system operation. The implemented IoT system successfully reduced irrigation time by 53.3% and water consumption by 41.6% compared to manual methods. Participant satisfaction reached 92%, with the highest ratings in the usefulness of the technology and the effectiveness of the training. These findings demonstrate that the application of IoT technology is effective in supporting plant maintenance and enhancing both productivity and the overall appeal of the Taman Inspirasi Strawberry agrotourism site.

Translating poetry is not about linguistic accuracy (Malmkjær, 2020), and if Torop finds that translation has not to be a mere word-by-word replacement (Torop, 2024), but has to deal with the transmission of whole experience of the original text, this is more the ever the case for poetry. The translation process is about getting the real voice and the real intention of the author, in prose, in drama and in poetry. It is about understating real cultural aspects, historical references and much more. Moreover, in poetry there is the challenge of the rhythm, metric and the rhetorical devices. AI has made big steps forward in the field of language and translation, but it is a fact that limitations persist in analysing and processing symbols, hidden messages and context, as well as – just to make an example – alliterations, assonances and metaphors. The aim of this study is to identify some of the challenges and main issues of Artificial Intelligence while translating poetry. Qualitative research will be conducted to analyse the role of AI in dealing with semantical, rhetorical and poetical aspects, so that translation choices do not prejudice any dimension. For this purpose, a contrastive analysis will be conducted between a Chat GPT-generated translation and a human-translated version of the same original text. This study focus on the poem Suicide in the Trenches by Siegfried Sassoon. Afterwards, the investigation will be about the analysis and interpretation of the poem, referring to the translation process into Italian itself. Considerations will be made in order to understand if AI was able to produce a translation which does need human intervention. Nevertheless, more empirical research is needed so that this study and its findings are generalised and positive and negative aspects in the translation process with AI are ultimately identified.

Artificial Intelligence (AI) is also transforming the judicial process because the application of tools has made the legal system more efficient, transparent, and accessible. Using AI-based predictive analytics, natural language processing, and chatbots, it is possible to simplify case management, conduct document reviews without human intervention, and conduct legal research more efficiently. The technologies will assist lawyers and judges working alongside other professionals in the courts in saving administrative costs and concentrating on making substantive decisions. Secondly, AI information analysis can help in detecting the pattern of cases, field of threat, and sentencing concerns, which can minimize the likelihood of the judicial decision being more uniform. Nevertheless, the greatest problems that emerge during the implementation of AI are justice, accountability, and the probability of algorithmic bias, which can disrupt judicial independence. The implementation thus needs to be controlled properly and under the control of human beings. Overall, AI can be a significant source of judicial modernization, although it must be applied conscientiously so that it is not just fair but novel.

From Handcrafted Features to Deep Learning: An Overview of Common Issues and Human Intervention in Object Detection

Artificial intelligence (AI) is a broad term that refers to the use of computers to replicate intelligent behavior with minimal human intervention. AI is rapidly transforming various sectors, including speech and language pathology, by offering innovative solutions to enhance therapeutic practices and client outcomes. Its application in speech and language pathology spans several domains, including medical diagnosis, therapeutic planning, and rehabilitation, utilizing tools such as machine learning and deep learning to enhance data analysis and pattern recognition. The primary aim of this study is to provide resources for speech and language pathologists on the topic of artificial intelligence by presenting research findings on the assessment and intervention of speech and language disorders using AI. Accordingly, AI studies in speech and language pathology found in the literature were included. The results of these studies were summarized, and information was provided on the use of AI in assessing and treating speech and language disorders, including swallowing disorders, voice disorders, acquired language disorders, motor and speech sound disorders, cleft palate speech, and developmental language disorder. Existing literature acknowledges and supports the growing popularity of AI and AI-based algorithms in speech and language pathology. Although the current evidence remains insufficient and concerns about ethics and implementation persist, advancing technology offers promise for applying AI in this field.

In this paper, a juvenile-adult population model incorporating seasonal succession and pulsed harvesting is developed. The seasonal succession captures the cyclical change between favorable and unfavorable environmental conditions, while the pulsed harvesting represents a periodic human intervention, targeting the adult population exclusively during favorable seasons. The principal eigenvalue for the corresponding linearized system is defined and its dependence on both the intensity of the harvesting pulses and the duration of the unfavorable season is analyzed. Explicit expressions and analysis of the principal eigenvalue for a logistic model extended with seasonal succession and pulsed harvesting are provided specifically. Based on the principal eigenvalue, we establish sufficient conditions for population persistence and extinction. Numerical simulations are conducted to validate these analytical results. Our findings demonstrate that higher harvesting intensity during the favorable season is detrimental to species survival. Furthermore, extending the duration of the unfavorable season can trigger a critical transition from population persistence to extinction.

Abstract Ensuring data integrity via Data Quality Monitoring (DQM) is critical in large-scale particle physics experiments. This has traditionally relied on labor-intensive manual inspection or static machine learning models, which are adequate for stable operating conditions. However, the current era of major detector upgrades at the Large Hadron Collider (LHC) challenges this paradigm. These upgrades are followed by prolonged commissioning periods with frequent and unpredictable changes in detector conditions, a regime for which static models are ill-suited.

To address this, we reframe DQM as a dynamic decision-making problem and introduce a human-in-the-loop reinforcement learning (RLHL) framework. Our Proximal Policy Optimization (PPO) agent learns both to classify data and to strategically decide when to query human experts, optimizing the automation-oversight balance. On synthetic data, the system rapidly adapts to abrupt condition changes and successfully learns from noisy labels. In a simulated online regime, the agent minimizes human intervention by requesting it mainly when its uncertainty is high.

A preliminary study on a real offline dataset from the LHCb experiment demonstrates that our synthetic approach is a reasonable proxy for real-world scenarios. The algorithm generalizes effectively with only superficial hyperparameter tuning, robustly identifying anomalies even when trained on augmented data. This work presents a scalable, adaptive solution for semi-autonomous DQM, paving the way for more intelligent control systems in complex scientific facilities.

Self-driving vehicle (SDV) safety and reliability are becoming critical design parameters as SDVs increase their market share. This paper examines public preferences for key SDV safety features (system reliability, sensor resilience, failure behavior, and driver alert methods) using a choice-based conjoint survey of 403 U.S. respondents. A novel integration of conjoint analysis with Least Absolute Shrinkage and Selection Operator (LASSO) regression and generalized linear mixed-effects models (GLMMs) was applied to identify the most influential features and their demographic or behavioral predictors. Results show that multimodal driver alerts (i.e., audio + visual) were the most influential factor, accounting for nearly two-thirds of decision weight. System reliability (i.e., low human intervention rates) and sensor resilience (i.e., low tolerance for failures) were secondary, while failure behavior had minimal influence. Subgroup analyses revealed modest variations by willingness to pay for SDVs, income, race/ethnicity, marital status, education, driving frequency, and risk propensity, though the importance of alerts and reliability remained consistent across groups. This combined conjoint-LASSO-GLMM framework enhances the precision of preference estimation and offers actionable guidance for SDV manufacturers seeking to align safety feature design with consumer expectations.

Abstract - In the modern era of data-driven decision-making, organizations rely heavily on clean, structured, and reliable data to achieve accurate analytical insights and machine learning outcomes. However, data wrangling, the process of cleaning, transforming, and enriching raw data, remains one of the most time-consuming and error-prone stages of data science. This research introduces LLama Wrangler, an intelligent and secure platform for automated data wrangling powered by Large Language Models (LLMs). The system simplifies the data preparation process by integrating artificial intelligence and cybersecurity to ensure both automation and data privacy. LLama Wrangler automates tasks such as data cleaning, feature type inference, data enrichment, and transformation using intelligent LLM-based algorithms. Furthermore, it embeds security mechanisms like encryption, access control, and privacy-preserving computation to handle sensitive data securely. By automating the wrangling pipeline, the system reduces human intervention by over 70%, minimizes errors, and enhances data integrity. Experimental results show that LLama Wrangler significantly improves data quality and model performance. This paper explores the motivation, methodology, architecture, evaluation, and future prospects of this innovative solution. Key Words: Data Wrangling, Large Language Models, Artificial Intelligence, Cybersecurity, Data Cleaning, Feature Type Inference, Data Enrichment, Machine Learning Automation.

The farming industry faces continuous threats from pest control and farm security issues because rodents cause significant damage to crops and disrupt farm operations. Traditional pest control methods require continuous human interaction which proves both resource-intensive and inefficient. Modern agricultural practices benefit from sustainable solutions through the combination of renewable energy with smart technologies. This research presents an innovative solar-powered motion-sensor system that utilizes OpenCV and YOLOv8 computer vision frameworks to autonomously detect and classify rodent intruders on farmland real-time. The system demonstrates its ability to detect and prevent rodent intruders according to initial testing results. The OpenCV/YOLO system uses motion sensor signals to analyze movement patterns before distinguishing rodents in their various groups. The solar-powered system operates continuously which decreases human intervention needs and enhances farm surveillance capabilities. The model demonstrates its capability to defend crops from rodent damage and enhance farm resistance against land degradation threats, resulting to improved crop yield and management. KPIs have been evaluable to prove the system’s efficiency and reliable to be used in agricultural practices to manage animal pests. The current system encounters problems with detecting wild animals beyond rodents as well as tracking rodent activity beneath ground level. Future developments could include improved pest capture systems alongside enhanced surveillance features for detecting both unauthorized human intruders and large animals. Future research should also invest in real-time implementation in fields with real data. The automated monitoring technology needs to be integrated with reliable sources of energy to create sustainable agricultural operations that are efficient and resilient, hence enhancing food security.

Purpose This paper investigates an integration of automation and digital technologies within air cargo logistics, focusing on empirical testing of the O3dyn pallet transport robot prototype at Munich Airport. The presented research identifies challenges and opportunities regarding dynamic airport environments, enhancing understanding of technology implementation and its implications for operational efficiency. Design/methodology/approach The conducted research employs an empirical methodology involving 10 days of real-world testing. This includes defining test parameters, documenting operational metrics and monitoring interactions toward other robotic systems. Various scenarios were tested to assess the system's effectiveness in navigating complex airport environments, collecting data on travel times, load weights, manual interventions and operational challenges relevant to air cargo handling. Findings The findings indicate that while the robot effectively performed transport tasks in a dynamic airport environment, its autonomy was limited, necessitating significant human intervention. Challenges included obstacle detection and navigation, indicating a need for further development in real-time decision-making and integration with logistics processes. Research limitations/implications The focus on a single airport may not fully capture broader challenges, and the short testing duration may overlook various operational scenarios. Future research should involve multiple and diverse environments and longer periods of data capturing for more comprehensive insights into air cargo handling systems. Practical implications This study provides guidance for air cargo logistics stakeholders, highlighting critical investment areas and the need for collaboration among industry partners to overcome automation barriers and challenges. Originality/value This paper presents unique empirical findings on air cargo robotics, demonstrating their practical implications and advancing the understanding of automation in dynamic airport environments.

Abstract This research is based on the idea that certain cognitive-intensive tasks typically carried out by structural engineers—such as choosing how to effectively arrange a building’s structure—can be successfully automated. In this article we investigate two techniques widely used in the field of Artificial Intelligence, namely Monte Carlo Tree Search (MCTS) and Genetic Algorithm (GA). Following a tabula rasa approach, according to which no hints nor external data are used as a reference for navigating the search space, we demonstrate how structural designs of two-dimensional multi-storey reinforced concrete structures can be generated, with no human intervention, by implementing and combining the two techniques from a reinforcement-learning perspective. The design tasks assigned to the developed software agents concern civil structures under static and seismic loads, and the basis for comparison is represented by a combination of construction cost and greenhouse gas emissions associated with the making of the structures. In the article, based on the main concepts of Computational Mechanics, a theoretical framework is introduced, which allows to represent both structures and design tasks in a simple, analytical way. The process of gamification, to which MCTS is often associated, is then described, so that structural design is reduced to the concepts of state, actions and payoff.. Finally, case studies are presented in which different agents—based respectively on GA, MCTS, and a combination of both—are tested. The results suggest that hybrid approaches, where GA operates first followed by MCTS, are the most effective.

Abstract. The generalized concept of "Human in the Loop" (HITL) enhances system performance by integrating human expertise into the decision-making process of agents. In a narrower sense, HITL specifically refers to human involvement in reinforcement learning (RL) through three key mechanisms: demonstration, intervention, and evaluation, each optimizing different stages of the training process. This approach effectively incorporates prior human knowledge, mitigates risks and sample bias in RL, and improves exploration efficiency and neural network convergence. However, existing HITL methods heavily rely on human experts for real-time annotations and guidance, leading to high implementation costs and operational complexity.In the domain of autonomous driving, traditional hierarchical decision-making frameworks depend on high-definition (HD) maps for planning and navigation. Notably, the construction of HD maps inherently embeds expert knowledge, semantic rules, and constraint information. Inspired by this observation, this study introduces an innovative approach: "HD Map in the Loop" (HMITL), leveraging HD map features as a substitute for human expertise and establishing a corresponding application framework for autonomous driving. Specifically, this research systematically investigates three core aspects of HMITL in training end-to-end decision-control models: (1) imitation learning based on expert demonstrations from HD maps; (2) Method for constructing action interference and reward function guided by HD map spatial heterogeneity; and (3) Critic priority architecture relying on expert evaluations from HD map perception and features. These three dimensions are logically interrelated and collectively form the foundational framework of HMITL. By pioneering this methodological innovation, this study provides a novel solution to reducing reliance on real-time human intervention in autonomous driving while ensuring the reliability and safety of system decision-making.

Writing formal emails requires clarity, conciseness, and accuracy. Many English as a Foreign Language (EFL) learners find these skills challenging. Large Language Models (LLMs), such as ChatGPT, represent a significant advancement in natural language processing. However, their effectiveness in structured, task-specific language tasks for education is not well studied. This study investigates whether the GPT-3.5 model can assist EFL students in writing formal emails with greater precision and structural accuracy. The research treats the classroom as a real-world test case for the model’s abilities. It applied a quasi-experimental method. Data were collected from control and experimental groups (N = 60). The study used a test and semi-structured interviews. The results show ChatGPT improved formal email writing among EFL students. Qualitative findings highlight the tool’s ease of use and time-saving nature. ChatGPT also boosts clarity and confidence. Some concerns about overreliance on ChatGPT emerge. The tool sometimes struggled with subject lines, sign-offs, and signatures. This suggests that human intervention is needed. Contributing to AI-assisted pedagogy, this study shows how ChatGPT can improve EFL students' formal email writing skills.

ChalanBeel, one of the major inland depressions of marshy character, is in the vulnerable condition due to various human interventions. The number of population of the ChalanBeel area has increased manifold in the recent years. To accommodate the increasing population, huge Beel areas were drained out and reclaimed by the locals. Roads, highways, embankments, bridges, culverts and other infrastructures were developed in this area. Due to increase of agricultural practices, the number of wetlands reduced over the period of time. Overfishing, pollution, unplanned infrastructures, lack of institutional coordination, lack of public awareness, etc. have become responsible for environmental degradation in this area. The present paper assesses the human intervention induced land use land cover changes in the ChalanBeel area. This study was conducted based on both primary and secondary data. Primary data was collected from field survey through interviewing the locals and the secondary data was collected from various published and online sources as well as satellite images were used to identify the land use land cover changes. The study reveals that built-up areas increased from 40.87 sq. km to 144.06 sq. km. and water bodies declined from 258.65 sq. km. to 90.63 sq. km in between 2003 and 2023, and the low-lying areas remained nearly same in that period. Land use land cover changes have significant impacts on the local people’s lives and livelihoods. It is important to reduce the human intervention induced land use land cover changes to ensure the planned growth of the surveyed areas. Jagannath University Journal of Life and Earth Sciences, 10(1) 40-58