Combination Of Strength Research Articles

Microstructure and mechanical properties of dissimilar resistance spot welds of TBF and HSLA300 steel sheets

Abstract The welding behavior of the resistance spot welded dissimilar TBF–HSLA300 steel joints under different welding parameters was investigated. TBF steels were used as galvanized or ungalvanized. Microstructural characterization and mechanical tests were performed. Higher heat input resulted in the higher possibility of shrinkage cavities at the weld nugget and also increased the risk of liquid metal embrittlement (LEM) crack in the heat affected zone (HAZ) on galvanized TBF steel sheets. A partial melting zone was observed on HSLA300 side. The welds with ungalvanized TBF sheets had the higher nugget size and indentation. The highest hardness values (over 500 HV0.1) were observed in HAZ on TBF side. The lowest hardness values (below 300 HV0.1) on TBF side were observed in a very narrow softening zone between tempered zone and base metal. The fractures of all spot welds in the tensile shear test occurred on the TBF steel side with the thinner thickness. Low-strength steel sheet have to be sufficiently thicker in order to achieve higher welding strength. Higher welding time led to higher weld strength in the ungalvanized TBF combination. In the galvanized TBF combination, higher heat input led to the drastic decrease in weld strength due to LEM cracks.

Predictive Modeling for Electric Vehicle Battery State of Health: A Comprehensive Literature Review

The rising adoption of electric vehicles (EVs) utilizing lithium-ion batteries necessitates a robust understanding of state-of-health (SOH) estimation. The existing literature highlights various SOH estimation models, but a comprehensive comparative analysis is lacking. This paper addresses this gap by conducting an exhaustive review of diverse SOH estimation approaches for EV battery applications, including the direct measurement method, physical-based and data-driven approaches. Results highlight that data-driven methods, particularly those utilizing machine learning techniques, offer superior accuracy and adaptability but often require extensive datasets. In contrast, physical-based approaches provide interpretable insights but are computationally intensive, and direct measurement methods, though simple, lack generalizability. In addition, this paper also systematically reviews the indicators of battery SOH, influential factors affecting battery SOH, and various datasets used for SOH modeling. Future research should focus on integrating multiple modeling methodologies to leverage their combined strengths, enhancing the collection of comprehensive battery lifecycle datasets to support robust model development, and extending the scope of SOH estimation beyond individual cells to encompass entire battery packs.

Positive Correlation Between Serum Limonene Levels and Muscle Health in a Representative Adult Population in the United States.

Background/Objectives: Monoterpenes, a class of organic compounds with the molecular formula C10H16, have garnered significant attention for their potential medicinal benefits. Emerging evidence suggests they may positively influence skeletal muscle function. However, the impact of monoterpene exposure on muscle strength and mass in humans remains unclear. Methods: To explore this relationship, we analyzed data from 1202 adults (aged ≥ 18 years) who participated in the 2013-2014 National Health and Nutrition Examination Survey (NHANES), focusing on serum levels of three specific monoterpenes-α-pinene, β-pinene, and limonene-and their association with hand grip strength and lean muscle mass. Results: Our analysis revealed that, except for test 2 of hand 1, all grip strength measures showed a positive correlation with ln-limonene levels. The β coefficient for combined grip strength was 2.409 (S.E. = 0.891, p = 0.015). Positive associations were also found between serum limonene levels and lean muscle mass. The β coefficient for the Appendicular Skeletal Muscle Mass Index (ASMI) was 0.138 (S.E. = 0.041, p = 0.004). Furthermore, combined grip strength and ASMI significantly increased across limonene quintiles (p for trend = 0.005 and 0.006, respectively). However, none of the three monoterpene levels showed a significant association with clinically defined low muscle mass or low muscle strength. Conclusions: Our findings suggest a plausible association between exposure to limonene, hand grip strength, and lean muscle mass among adults in the United States. Further investigation is needed to fully understand the underlying mechanisms and medical significance of this association.

SynthSecureNet: An Improved Deep Learning Architecture with Application to Intelligent Violence Detection

We present a new deep learning architecture, named SynthSecureNet, which hybridizes two popular architectures: MobileNetV2 and ResNetV2. The latter have been shown to be promising in violence detection. The aim of our architecture is to harness the combined strengths of the two known methods for improved accuracy. First, we leverage the pre-trained weights of MobileNetV2 and ResNet50V2 to initialize the network. Next, we fine-tune the network by training it on a dataset of labeled surveillance videos, with a focus on optimizing the fusion process between the two architectures. Experimental results demonstrate a significant improvement in accuracy compared with individual models. MobileNetV2 achieves an accuracy of 90%, while ResNet50V2 achieves a 94% accuracy in violence detection tasks. SynthSecureNet achieves an accuracy of 99.22%, surpassing the performance of individual models. The integration of MobileNetV2 and ResNet50V2 in SynthSecureNet offers a comprehensive solution that addresses the limitations of the existing architectures, paving the way for more effective surveillance and crime prevention strategies.

Collaborative Healthcare Data Management Framework using Parallel Computing and the Internet of Things

Healthcare data management has become a critical research area, primarily driven by the widespread adoption of personal health monitoring systems and applications. These systems generate an immense volume of data, necessitating efficient and reliable management solutions for lossless sharing. This article introduces a Collaborative Data Management Framework (CDMF) that leverages the combined strengths of parallel computing and federated learning. The proposed CDMF is designed to achieve two primary objectives: reducing computational complexity in data handling and ensuring high sharing accuracy, regardless of the data generation rate. The framework employs parallel computing to streamline the scheduling and processing of data acquired at various intervals. This approach minimizes processing delays by operating on a less complex scheduling algorithm, making it suitable for handling high-frequency data generation. Federated learning, on the other hand, plays a pivotal role in verifying data distribution and maintaining sharing accuracy. By enabling decentralized learning, federated learning ensures that data remains on local devices while sharing only the necessary model updates. This approach enhances privacy and security, a critical consideration in healthcare data management. It ensures that data distribution and sharing are verified based on appropriate requests while avoiding latency issues. By decentralizing the learning process, federated learning enhances privacy and security, as raw data does not leave the local systems. This cooperative interaction between parallel computing and federated learning operates in a cyclic manner, allowing the framework to adapt dynamically to increasing monitoring intervals and varying data rates. The performance of the CDMF is validated through improvements in two key metrics. First, the framework achieves a 15.08% enhancement in sharing accuracy, which is vital for maintaining data integrity and reliability during transfers. Second, it reduces computation complexity by 9.48%, even when handling maximum data rates. These results highlight the framework’s potential to revolutionize healthcare data management by addressing the dual challenges of scalability and accuracy.

Effect of Thermal Ageing on the Electrical Insulation Properties of the Oil-Paper Insulation

The article focuses on the electrical insulation properties of oil-paper insulation and its application in practice at AC voltage. It also discusses the effect of moisture and thermal ageing on the breakdown strength and breakdown voltage of oil-paper combinations. Measurements are made of the breakdown voltage and breakdown strength of liquid insulators and paper type KREMPEL-DPP, impregnated with two types of oils, SHELL Diala S4 ZX-I and Mogul Trafo CZ-A. The moisture content in the liquid insulants before and after thermal ageing is also measured.

Exercise Intervention in Women with Fibromyalgia and Its Influence on Pain, Psychological Variables, and Disability: An Observational Study.

(1) Background: Fibromyalgia syndrome (FM) is a specific condition within the spectrum of musculoskeletal pain disorders, with an estimated global prevalence of 2%. Physical exercise has shown promise in modulating pain and improving physical function without the drawbacks of pharmacotherapy. This study aims to examine the effects of a 6-week telerehabilitation combined exercise program-including mobility, strength, and high-intensity exercises-on pain, psychological variables, and disability in women with fibromyalgia. (2) Methods: In this observational study involving 53 FM patients, the outcomes measured were the pressure pain threshold (PPT), the conditioned pain modulation (CPM) paradigm, levels of pain on the measurement day and the average of the last week (using NRS) the impact of the fibromyalgia (using Fibromyalgia Impact Questionnaire (FIQ), and anxiety (using the Spanish version of the State-Trait Anxiety Inventory-STAI). (3) Results: Statistically significant changes were observed in the intervention group in PPT, CPM, NRS, and FIQ. (4) Conclusions: A six-week telerehabilitation therapeutic exercise intervention consisting of two scheduled exercise sessions per week lasting approximately 45 min each is associated with reduced pain levels, enhanced pain inhibitory pathways, and a decreased impact of fibromyalgia compared to patients who do not adopt a more active lifestyle.

A Novel Network Optimization Framework Based on Software-Defined Networking (SDN) and Deep Learning (DL) Approach

Access to networks and the Internet has multiplied, and data traffic is growing exponentially and quickly. High network utilization, along with varied traffic types in the network, poses a considerable challenge and impact on the ICT Infrastructure, particularly affecting the performance and responsiveness of real-time application users who will experience slowness and poor performance. Conventional/traditional Quality of Service (QoS) mechanisms, designed to ensure reliable and efficient data transmission, are increasingly insufficient due to their static nature and inability to adapt to the dynamic demands of modern networks. As such, this study introduces a Novel Network Optimization Framework leveraging the combined strengths of Software-Defined Networking (SDN) and Deep Learning (DL) to dynamically manage multiple QoS of network devices in enterprise and campus network environments. The proposed system is a dynamic QoS that utilizes SDN's global monitoring and centralized management control capabilities to programmatically control network devices, ensuring that sensitive traffic is allocated with appropriate bandwidth and minimized latency. Concurrently, DL algorithms enhance the framework's decision-making process by proposing an accurate preferred configuration for the best adequate bandwidth for sensitive traffic transmission. This integration facilitates real-time adjustments to network conditions and improves overall network performance by ensuring high-priority applications receive the bandwidth they require without manual/human intervention. By providing a dynamic, intelligent solution to QoS management, this framework represents a significant step forward in developing more adaptable, resilient, and efficient networks capable of supporting the demands of contemporary and future digital ecosystems.

Improving health outcomes through a strength training and yoga program for young women with PCOS: A quality improvement project

Background: Polycystic Ovary Syndrome (PCOS) is a common endocrine disorder associated with physical, mental, and reproductive health challenges. Lifestyle interventions, including exercise and yoga, have shown promise in managing PCOS symptoms. This quality improvement project aimed to assess the impact of a combined strength training and yoga program on the health outcomes of young women with PCOS. Methods: The 13-month program (December 2021 to January 2023) was conducted in a community setting. Participants were young women diagnosed with PCOS, selected based on predefined inclusion criteria. The intervention included bi-weekly strength training and yoga sessions, supplemented by group discussions and motivational coaching. Primary outcomes included weight loss, BMI reduction, mental health improvement (measured by PHQ-9 and GAD-7 scores), and menstrual regularity. Adherence and participation rates were tracked to evaluate engagement. Results: The average weight loss among participants was 5.2 ± 2.4 kg, with BMI reduced from 29.1 ± 3.2 kg/m² to 26.8 ± 2.9 kg/m². PHQ-9 depression scores decreased from 12.3 ± 4.5 to 6.4 ± 3.2, and GAD-7 anxiety scores improved from 10.6 ± 3.7 to 5.8 ± 2.9. Menstrual regularity improved, with 75% of participants reporting regular cycles by the end of the program. 82% of participants completed at least 70% of the sessions, indicating strong adherence. Conclusion: The combined strength training and yoga program effectively improved physical, mental, and reproductive health outcomes among young women with PCOS. High adherence rates demonstrated the feasibility of this approach, emphasizing the importance of lifestyle interventions in managing PCOS. Future efforts should focus on scaling the program and integrating nutrition-based strategies for even better outcomes.

Electronic Health Records-Based Data-Driven Diabetes Knowledge Unveiling and Risk Prognosis

In the healthcare sector, the application of deep learning technologies has revolutionized data analysis and disease forecasting. This is particularly evident in the field of diabetes, where the deep analysis of Electronic Health Records (EHR) has unlocked new opportunities for early detection and effective intervention strategies. Our research presents an innovative model that synergizes the capabilities of Bidirectional Long Short-Term Memory Networks-Conditional Random Field (BiLSTM-CRF) with a fusion of XGBoost and Logistic Regression. This model is designed to enhance the accuracy of diabetes risk prediction by conducting an in-depth analysis of electronic medical records data. The first phase of our approach involves employing BiLSTM-CRF to delve into the temporal characteristics and latent patterns present in EHR data. This method effectively uncovers the progression trends of diabetes, which are often hidden in the complex data structures of medical records. The second phase leverages the combined strength of XGBoost and Logistic Regression to classify these extracted features and evaluate associated risks. This dual approach facilitates a more nuanced and precise prediction of diabetes, outperforming traditional models, particularly in handling multifaceted and nonlinear medical datasets. Our research demonstrates a notable advancement in diabetes prediction over traditional methods, showcasing the effectiveness of our combined BiLSTM-CRF, XGBoost, and Logistic Regression model. This study highlights the value of data-driven strategies in clinical decision-making, equipping healthcare professionals with precise tools for early detection and intervention. By enabling personalized treatment and timely care, our approach signifies progress in incorporating advanced analytics in healthcare, potentially improving outcomes for diabetes and other chronic conditions.

Random shear resistance of a headed-stud connector in composite steel-concrete beam

In the traditional, standard, computational approach, the shear resistance of a single headed-stud connector, ensuring the composite connection between a steel beam and a reinforced concrete slab resting on this beam, is determined by comparing the load capacity PRs – determined by the destruction of the steel connector itself, and the load capacity PRc – conditioned by the destruction of the concrete surrounding this connector. In a single implementation, the smaller of these both values, i.e. PR = min (PRs, PRc ), is authoritative for the designer. If, however, both combined strengths are treated as the random variables and a statistically homogeneous sample grouping potentially possible implementations of this type is taken into account, then the design resistance PR;d = [min (PRs, PRc )]d, representative for the verification of ultimate limit state for the considered connection, will be quantitatively different from the value PR,d = [min (PRs,d, PRc,d ) recommended for use in this regard in the standard EN 1994-1-1. In this paper a detailed algorithm for the correct specification of this value is presented in detail. The dependence of such value on the mutual relationship between the coefficients determining the statistical variability of the strength of the connector steel as well as the strength of the concrete from which the floor slab was made is also demonstrated. The proposed approach is based on the fully probabilistic design format, according to which the appropriate level of the probability of reliable work of the analyzed connection is ensured. The presented considerations are illustrated with a numerical example. On its basis, the degree of simplification of such evaluation is estimated, as well as its consequences, resulting from the use of a conventional standard model in this respect.

Research on an Elderly Indoor Fall Detection System Based on IoT and Computer Vision Technology: Integration and Performance Analysis of YOLOv5 and OpenPose

The rapid growth of the global elderly population has led to an increased need for efficient and reliable fall detection systems to ensure timely medical assistance. Traditional monitoring methods, such as wearable devices and environmental sensors, often face challenges in accuracy, reliability, and user compliance. This study proposes an indoor fall detection system for the elderly, integrating Internet of Things (IoT) and computer vision technologies, specifically utilizing YOLOv5 and OpenPose algorithms. YOLOv5 is employed for rapid and accurate human body detection, providing position information by computing the center of objects in frames or videos. OpenPose is then used for detailed real-time pose estimation, detecting 135 key points on the human body—including hands, face, and feet—to assess whether a fall has occurred based on posture analysis. The system was tested using two datasets, GMDCSA and URFD, achieving sensitivity values of 0.9412 and 0.9583, respectively, indicating high accuracy in fall detection. The false positive rates were 0.0588 and 0.2857, while the false negative rates were 0.0588 and 0.0417, demonstrating the system's reliability in minimizing detection errors. The integration of YOLOv5 and OpenPose leverages their combined strengths in object detection and pose estimation, resulting in a robust solution suitable for real-time applications.

Directed Forgetting and the Production Effect.

The item-based directed-forgetting effect is explained as a difference in how strongly people encode remember-cued over forget-cued targets. In contrast, the production effect is typically explained as a difference in the distinctiveness of the memory of produced over unproduced targets. The procedural alignment of the two effects - directing participants to remember or forget, produce or not - coupled with their different theoretical explanations (i.e., strength vs. distinctiveness) presents an opportunity to investigate common versus differential effects of elaborative encoding. This study aims to bridge the gap between these two well-established phenomena by comparing the differences in directed forgetting and the production effect in the context of recognition. Mixed- and pure-list designs were utilized to provide an index of each of these mechanisms in both procedures. Along with a standard production effect and directed forgetting effect in the mixed-list conditions, we found evidence for strength primarily driving results in both procedures. Results are explained using a global matching model of recognition memory, MINERVA 2, by assuming varying levels of encoding strength in relation to task demands. Critically, we obtain the best fit using a strength mechanism over a combined strength and distinctiveness mechanism for our data.

A hierarchical Bayesian brain parcellation framework for fusion of functional imaging datasets

Abstract Different task-based and resting-state imaging datasets provide complementary information about the organization of the human brain. Brain parcellations based on single datasets will therefore be biased towards the particular type of information present in each dataset. To overcome this limitation, we propose here a hierarchical Bayesian framework that can learn a probabilistic brain parcellation across numerous task-based and resting-state datasets, exploiting their combined strengths. The framework is partitioned into a spatial arrangement model that defines the probability of each voxel belonging to a specific parcel (the probabilistic group atlas), and a set of dataset-specific emission models that define the probability of the observed data given the parcel of the voxel. Using the human cerebellum as an example, we show that the framework optimally combines information from different datasets to achieve a new population-based atlas that outperforms atlases based on single datasets. Furthermore, we demonstrate that using only 10 min of individual data, the framework is able to generate individual brain parcellations that outperform group atlases.

Automatic grading diabetic retinopathy in color fundus image: Cascaded hybrid attention network

Deep learning techniques have been introduced to automatically assess Diabetic Retinopathy (DR), a prevalent vision-impairing condition, aiding ophthalmologists in formulating personalized treatment strategies for patients. However, most Artificial Intelligence (AI) for grading DR are constrained by their limited capacity to discern lesion details or necessitate manual lesion labeling, leading to suboptimal assessment outcomes or increased workload. To enhance both efficiency and precision in grading, this paper proposes a Cascaded Hybrid Attention Network (CHAN) for DR grading. Residual Hybrid Attention Modules (RHAM) is designed to extract features, incorporating several cascaded Hybrid Attention Module (HAM) and a convolution layer with a residual connection. The approach integrates channel attention with multi-head self-attention, leveraging their combined strengths: global statistical analysis and robust local adaptation. The grading score is derived by integrating DR features across various layers, from shallow to deep. The CHAN model, tested on the MESSIDOR dataset, demonstrates superior accuracy in DR grading, consistently surpassing current leading methods across evaluation metrics.

Plasma omega-3 fatty acids are not associated with muscle strength, regardless of age or protein intake: National Health and Nutrition Examination Survey 2011-12.

Muscle strength is a key predictor of both quality of life and mortality. Although numerous studies have investigated the relationship between omega-3 (ω-3) intake and muscle strength, the evidence remains inconclusive. Furthermore, it is unclear whether this association is influenced by protein intake. Therefore, the aim of this study was to evaluate the potential associations between plasma ω-3 levels (as a biomarker of ω-3 intake) and muscle strength across different age ranges and levels of protein intake. A cross-sectional study was conducted involving 1708 individuals aged 20-80 years from the 2011-2012 National Health and Nutrition Examination Survey (NHANES). Plasma levels of total ω-3 fatty acids and their subtypes-alpha-linolenic acid (ALA), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA)-were measured using chromatography-mass spectrometry. Muscle strength was assessed using handgrip strength, with the combined grip strength of both hands being analyzed. Dietary intake was evaluated using two 24-h recalls, and usual intake was estimated using the National Cancer Institute method. Linear regression analyses were conducted to examine the associations between plasma ω-3 fatty acids and muscle strength, stratified by age groups (<60 years and ≥60 years) and protein intake categories (<0.8g/kg/day, ≥0.8-<1.2g/kg/day, and ≥1.2g/kg/day), while adjusting for potential confounders. Total plasma ω-3, ALA, EPA, and DHA were not associated with muscle strength in the total sample, regardless of protein intake levels. This lack of association remained consistent across all age groups and protein intake strata. Plasma ω-3 fatty acids were not associated with muscle strength, regardless of age or protein intake. Overall, these findings suggest that plasma ω-3 fatty acids may not play a significant role in muscle strength.

Design of 3D printed concrete masonry for wall structures: Mechanical behavior and strength calculation methods under various loads

The advancement of 3D printed concrete (3DPC) technology for large-scale construction is hindered by inadequate load-bearing design methods for 3DPC wall structures. This work focuses on 3D printed concrete masonry (3DPCM) walls and aims to investigate the mechanical behavior and methods for computing the strength of 3DPCMs under various sectional forms and loads. Results showed that increasing the number of inclined ribs significantly enhances the shear strength without affecting the compressive strength. The combined shearcompression strength exhibits a nonlinear relationship with the axial compression ratio, reaching its maximum at a ratio of 0.4. The failure modes of 3DPCM under various loads were revealed through experiments and numerical simulations. The oblate spheroidal pores, with major axes aligned along the X-axis printing direction at intralayer interfaces, were identified as critical to 3DPCM structural failure. Finally, a strength calculation method for the compressive, shear, and combined shearcompression of the 3DPCM was proposed.

Machines think, but do we? Embracing AI in flat organizations: Challenges of technological change in women-owned manufacturing firms

Drawing on the combined strengths of Conservation of Resources (COR) theory and Information Processing Theory (IPT), we explored the relationship between technological turbulence and knowledge hiding, with employees’ awareness of AI as a key mediating factor. We also examined the dual moderating effects of hierarchical flatness in an organization and change leadership to test the theoretical synergy of COR and IPT. Data were collected from 315 women-owned manufacturing firms and analyzed through PLS-SEM. The findings reveal that employees’ AI awareness mediates the link between technological turbulence and knowledge-hiding behaviors. A flatter organizational structure aids in spreading AI awareness among employees, while effective change leadership serves as a crucial moderator, reducing knowledge hiding by influencing employee behavior. The survey highlights that increased AI awareness can lead to fears about AI's impact on job security, driving knowledge hiding. However, skilled leadership can mitigate this by promoting innovation and encouraging collaboration with external partners. This study provides original insights by developing and empirically validating a model of knowledge hiding in response to technological turbulence in women-owned manufacturing sectors in Punjab, Pakistan. It deepens the understanding of knowledge management by examining the interplay between organizational flatness and AI integration, areas often overlooked in existing literature.

A fine-tuning enhanced RAG system with quantized influence measure as AI judge

This study presents an innovative enhancement to retrieval-augmented generation (RAG) systems by seamlessly integrating fine-tuned large language models (LLMs) with vector databases. This integration capitalizes on the combined strengths of structured data retrieval and the nuanced comprehension provided by advanced LLMs. Central to our approach are the LoRA and QLoRA methodologies, which stand at the forefront of model refinement through parameter-efficient fine-tuning and memory optimization. A novel feature of our research is the incorporation of user feedback directly into the training process, ensuring the model’s continuous adaptation to user expectations and thus, improving its performance and applicability. Additionally, we introduce a Quantized Influence Measure (QIM) as an innovative “AI Judge” mechanism to enhance the precision of result selection, further refining the system’s accuracy. Accompanied by an executive diagram and a detailed algorithm for fine-tuning QLoRA, our work provides a comprehensive framework for implementing these advancements within chatbot technologies. This research contributes significant insights into LLM optimization for specific uses and heralds new directions for further development in retrieval-augmented models. Through extensive experimentation and analysis, our findings lay a robust foundation for future advancements in chatbot technology and retrieval systems, marking a significant step forward in the creation of more sophisticated, precise, and user-centric conversational AI systems. We make the dataset, the data processing package huggify-data, the model, and the app publicly available for the community.

HiMolformer: Integrating graph and sequence representations for predicting liver microsome stability with SMILES

In the initial stages of drug discovery or pre-clinical studies, understanding the metabolic stability of new molecules is crucial. Recently, research on pre-trained deep learning for molecular property prediction has been actively progressing, with various models being made open-source. However, most of these models rely on either 2D graph or 1D sequence for training, and the representation varies depending on the data format used. Consequently, combining multiple representations can broaden the scope of learning and may potentially be a manageable and most effective method to enhance performance.Therefore, we propose a novel hybrid model for predicting metabolic stability, which integrates representations from both graph-based and sequence-based models pre-trained for molecular features. This approach utilizes the combined strengths of 2D topological and 1D sequential information of molecules. HiMol, a graph-based graph neural network (GNN) model, and Molformer, a sequence-based Transformer model, were selected for integration, thus we named it HiMolformer. HiMolformer demonstrated superior performance compared to other models. We also focus on regression task for prediction with a empirical dataset from Korea Chemical Bank (KCB), comprising 3,498 molecules with mouse liver microsome (MLM) and human liver microsome (HLM) data obtained from actual metabolic reaction experiments. To the best of our knowledge, it is the first attempt to develop MLM and HLM prediction models using regression with a single SMILES input. The source code of this model is available at https://github.com/YUNSEOKWOO/HiMolformer.