Predicting Human Tactile Smoothness/Roughness Perception from Multidimensional Mechanical Properties of Synthetic Fibers Using Machine Learning

Predicting Human Tactile Smoothness/Roughness Perception from Multidimensional Mechanical Properties of Synthetic Fibers Using Machine Learning | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Predicting Human Tactile Smoothness/Roughness Perception from Multidimensional Mechanical Properties of Synthetic Fibers Using Machine Learning Hyung-Tak Lee, Jun-Yeop Kim, Keungyonh Bak, Kiun KIM, Sungwoo Chun, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7046348/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 27 Nov, 2025 Read the published version in Scientific Reports → Version 1 posted 13 You are reading this latest preprint version Abstract Accurately predicting human perception of tactile roughness remains challenging because previous models often used limited mechanical properties, small sample sizes, and insufficient validation methods. To address these limitations, we developed a predictive model integrating multidimensional mechanical properties and subjective evaluations of tactile perception, using 50 commercially available synthetic fiber samples, including polyester, spandex, nylon, and their blends. Twelve mechanical properties were measured across four categories: geometric roughness, frictional force, hardness, and tensile strength. Tactile perception of smoothness/roughness was evaluated by 37 participants using a 5-point scale, with lower values indicating smoother textures and higher values indicating rougher textures. Correlation analysis identified kinetic friction coefficient (KF, ρ = -0.67), arithmetic mean roughness (Ra, ρ = 0.44), mean width of profile elements (RSm, ρ = 0.42), maximum load (ML, ρ = -0.41), and root mean square slope (Rdq, ρ = 0.31) as key predictors. Among six regression models, Gaussian process regression showed the highest predictive accuracy (cross-validated R² = 0.71). Comparisons between non-cross-validated and cross-validated results revealed substantial performance drops in cross-validation, underscoring the risk of performance overestimation without rigorous validation. The proposed framework provides a robust, generalizable approach applicable to broader tactile dimensions, benefiting material evaluation, product development, and haptic technologies. Physical sciences/Engineering Physical sciences/Materials science tactile perception smoothness/roughness perception mechanical properties regression model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Introduction Tactile perception refers to the subjective sensation formed by integrating information received through receptors distributed across the skin. This perception includes tactile attributes such as smoothness/roughness, softness/hardness, coldness/warmth, and springiness/stiffness. These tactile attributes play essential roles not only in evaluating material comfort, but also in shaping user experience across a wide range of domains, including consumer products, virtual reality, robotics, and rehabilitation 1 – 3 . Smoothness/roughness is particularly influential in how users judge surface quality and usability in both daily interactions and specialized applications. Therefore, developing reliable methods for evaluating smoothness/roughness is crucial for quality control, material development, and consumer satisfaction across industries. Given this importance, researchers have proposed various quantitative methods to evaluate smoothness/roughness based on surface mechanical properties, ranging from surface friction measurements to advanced analyses of nanoscale microstructures. Recent studies have achieved remarkable spatial resolutions (10 nm–1 µm) using sophisticated measurement techniques for surface smoothness/roughness measurement 4 , 5 . However, these quantitative mechanical measurements alone cannot fully capture human perceptions of smoothness/roughness 6 , 7 because such perception involves subjective interpretations influenced by multisensory integration, context, and individual variability. Consequently, integrating quantitative mechanical measurements with subjective evaluations of smoothness/roughness is increasingly recognized as essential. Early studies primarily explored relationships between perceived smoothness or roughness and individual mechanical properties, such as surface geometry or frictional force. More recent research emphasizes that tactile perception arises from the complex interactions among multiple mechanical factors, including surface topography, height variation, spacing patterns, and material composition 8 . Multiparameter modeling approaches have consistently outperformed single-feature methods, showing significant improvements in predictive performance (R²) of approximately 0.20 to 0.35 8,9 . These findings highlight the necessity of multidimensional approaches that comprehensively integrate geometric properties (e.g., surface height and spacing), contact-related properties (e.g., friction and hardness), and intrinsic material properties for accurate tactile perception modeling. While multidimensional approaches are essential, achieving a balance between experimental control and practical relevance is equally important. Most previous studies used artificially manufactured samples via 3D printing or lithography to ensure precise control. Although these studies reported high predictive accuracy (classification accuracies and regression R² values ranging from approximately 0.75 to 0.89), their generalizability to real-world materials, influenced by multiple uncontrolled factors, remains limited 10 – 12 . Artificial samples typically simplify complex real-world textures by varying only a few parameters, insufficiently capturing the intricate interplay among mechanical factors. To overcome this limitation, it is necessary to investigate perception using realistic materials that naturally embody diverse mechanical interactions. As research increasingly emphasizes linking mechanical properties to human tactile perception using real-world samples, developing predictive models with practical generalizability has become crucial. Achieving this involves multidimensional feature integration coupled with rigorous validation. Cross-validation, particularly, ensures models accurately predict tactile responses for unseen materials. Recent studies by Lee et al. and James et al. illustrated that neglecting proper cross-validation procedures leads to inflated accuracy, dramatically reducing predictive reliability on new data 13 , 14 . Although previous tactile studies reported high performance (R² values of 0.72–0.92), many evaluated their models using the same datasets used for training, limiting their applicability 10 , 12 , 15 . To ensure reliable and transferable outcomes, predictive tactile models must incorporate robust cross-validation procedures. This study aims to develop a predictive model for human smoothness/roughness perception based on multiple mechanical properties of fiber materials, making three significant contributions to tactile perception research. First, we systematically selected nine representative mechanical features from an initial set of twelve through multicollinearity analysis, enabling comprehensive yet non-redundant characterization of fiber surfaces. This approach represents a methodological improvement compared to earlier studies that typically considered fewer parameters without systematic feature selection. Second, we used 50 diverse synthetic fibers commonly encountered in daily life, including polyester, spandex, nylon, and their composites. This sample size significantly expands upon the relatively small number of samples (typically 10–20) used in earlier research. Third, we rigorously evaluated model performance through five-fold cross-validation, addressing previous limitations of studies reporting high predictive accuracy without independent validation. Utilizing these selected features, we developed and evaluated six regression models, determining the most effective approach for predicting smoothness/roughness perception. Our approach provides a robust and transferable framework for modeling tactile perception, with potential applications in material evaluation, product development, and haptic technology design. Furthermore, the established framework has the potential to be extended to other tactile dimensions such as softness, hardness, and thermal sensations. Methods Material Preparation In this study, 50 synthetic fibers commonly used in practical applications were selected for experimental analysis. The fiber samples were classified into single-component and composite groups. The single-component group consisted of 30 samples of 100% polyester and 2 samples of 100% nylon. The composite group included 18 blended samples: polyester-based (9 polyester/spandex, 2 polyester/polyurethane combinations) and nylon-based blends (5 nylon/spandex, 2 nylon/polyurethane combinations). These fiber types were chosen to reflect materials frequently encountered in textile and apparel products, enabling the development of a practically relevant model for predicting tactile smoothness/roughness perception. Polyester, representing approximately 70% of the global fiber market, was primarily selected due to its significant market share, durability, cost-effectiveness, and widespread industrial and consumer applications 16 . Nylon and synthetic blends were also included because of their common use in functional and everyday clothing, offering diverse tactile and mechanical characteristics such as variations in friction, elasticity, and surface texture. The inclusion of blended materials particularly enhances the model's applicability, reflecting the complexity of real-world tactile experiences. Mechanical Property Measurement Human tactile smoothness/roughness perception arises from the integration of various stimuli detected by mechanoreceptors in the skin during surface contact. To accurately model this complex perceptual process, we measured 12 mechanical properties across four categories that comprehensively characterize fiber surfaces: First, two frictional properties of the samples were assessed using standardized testing methods. Although these tests may not fully replicate finger-surface interactions, they allow objective comparisons among materials and have been validated as reliable predictors of tactile perception in numerous studies 17 . Second, geometric roughness properties were evaluated through seven parameters defined in the ISO 4287 standard. Third, Shore hardness was measured to account for the influence of material hardness on tactile sensation. Hardness has been demonstrated by Yeo et al. (2017) to significantly impact tactile feedback during contact 18 . Lastly, two tensile properties, including strength and extensibility, were examined to reflect material deformation under tactile interaction forces. These mechanical attributes significantly contribute to tactile experiences in textile perception 19 . The twelve selected properties collectively represent diverse aspects of tactile interactions between human skin and fiber surfaces. This multidimensional approach effectively captures the complex microstructure characteristics that cannot be adequately represented by a single roughness parameter 20 . Each property was measured 10 times per sample, and mean values were calculated to ensure measurement reliability. Friction Force Measurement The frictional characteristics of the fiber samples were quantified by measuring the static (SF) and kinetic friction (KF) coefficients using a friction coefficient tester (QMESYS, Korea) 21 . Measurement conditions included a load cell of 1.0 kgf, movement speed of 200.0 mm/min, and friction weight of 200.0 g. Friction force was measured by securing each fiber sample to the friction element. The SF coefficient was recorded at the initiation of movement, while the KF coefficient was measured during constant velocity motion. Geometric Smoothness/Roughness Measurement Geometric smoothness/roughness of the fiber samples was measured using an Alpha-Step IQ surface profiler (KLA Tencor, USA) 22 . Measurement parameters included a sampling rate of 200 Hz, scan length of 5.0 mm, and scan speed of 0.1 mm/sec. Seven parameters were extracted from the profiles based on ISO 4287 standard: arithmetic mean roughness (Ra), root mean square roughness (Rq), maximum height of the profile (Rz), mean width of the profile elements (RSm), skewness of the profile (Rsk), kurtosis of the profile (Rku), and root mean square slope (Rdq). Hardness Measurement Shore hardness (SH) of fiber samples was measured using an HT-6510 hardness testers (REED Instruments, USA) 23 . The tester was mounted on a support stand, and SH was measured at multiple points on each fiber sample to obtain representative hardness values. Tensile Strength Measurement Maximum load (ML) and elongation at break (EB) of the fiber samples were measured using Instron tensile testing machine (Instron, USA) 24 . Measurements were conducted at a tensile speed of 100.0 mm/min. ML represents the peak force that fibers withstand before breaking, while EB is expressed as the percentage of extended length relative to initial length at rupture, quantitatively evaluating the strength and extensibility of each fiber sample. Tactile Perception Evaluation Participants The experiment was conducted with 37 healthy adult volunteers (23 males and 14 females; mean age: 24.1 ± 2.6 years). Participants had no mental or physical disorders, nor any history of sensory nerve damage. The study protocol was approved by the Institutional Review Board of Korea University (IRB-2024-0163). All processes were conducted in accordance with the relevant guidelines and regulations, including the principles outlined in the Declaration of Helsinki. Before the experiment, participants were informed about the study objectives, procedures, and necessary precautions, after which they provided informed consent. Appropriate compensation was provided upon completion of the experiment. Experimental Procedure The tactile smoothness/roughness perception experiment was conducted using 50 fiber samples. Participants' hands were thoroughly cleaned prior to evaluation to eliminate interference from foreign substances or perspiration. Detailed instructions on the evaluation method were provided, and participants practiced with sample materials until they were comfortable with the procedure. All fiber samples were stored under identical environmental conditions and presented in random order. To minimize perceptual variations, tactile perception was assessed using only the distal phalanx of the right index finger 25 . Participants moved their index finger horizontally across each fiber sample, ensuring consistent tactile stimulation and reliable assessments. Participants classified each sample on a 5-level scale, with lower values indicating smoother textures and higher values indicating rougher textures. This 5-point classification scheme is consistent with established practices in tactile research, effectively capturing meaningful differences in human tactile perception with optimal sensitivity and minimal complexity [31–33]. The use of this approach ensured a balanced distribution of samples across perception levels (Fig. 1 ), providing adequate statistical power and reducing potential bias during regression analyses, particularly within the cross-validation procedure for assessing predictive model generalizability. The responses from all 37 participants were averaged for each fiber sample to derive representative tactile perception values, which were subsequently used for modeling the correlation between human tactile perception and physical properties. Data Analysis and Modeling Data Preprocessing The mechanical properties possess varying units and ranges, complicating their direct comparison in smoothness/roughness perception prediction models. For instance, Ra is measured in micrometers (µm), two friction coefficients (SF and KF) are dimensionless values, and SH uses a scale from 0-100. Such scale discrepancies may cause disproportionate influence of certain features or underestimation of important ones during model training. To resolve this issue, each mechanical property was normalized using standard scores (z-scores) 26 , calculated as follows: $$\:\begin{array}{c}Z\:=\:\frac{\text{X}\:-\:{\mu\:}\:}{{\sigma\:}}\#\left(1\right)\end{array}$$ where Z is the standardized value, X represents the original data value, µ is the mean of the respective feature, and σ is the standard deviation of the feature. This normalization transforms all mechanical properties into distributions with a mean of 0 and a standard deviation of 1, facilitating direct feature comparison. Normalized data enhances the accuracy of feature importance assessment and improves the convergence and stability of machine learning algorithms. Multicollinearity Analysis Addressing multicollinearity among independent variables is essential for accurate and reliable regression modeling. Multicollinearity occurs when independent variables exhibit high correlations, potentially destabilizing regression coefficients and complicating model interpretation. Specifically, including highly correlated variables together makes it difficult to assess their distinct effects, which can negatively impact model performance 17 . To identify multicollinearity issues, Variance Inflation Factor (VIF) analysis was conducted on the initial 12 mechanical properties. A VIF value exceeding 10 generally indicates severe multicollinearity [27]. The analysis identified severe multicollinearity in SF (153.16), KF (157.19), Rz (217.62), Ra (1151.40), and Rq (2075.50), while acceptable VIF values were observed for Rsk (1.57), Rdq (2.53), RSm (2.78), Rku (4.32), SH (1.67), EB (1.44), and ML (1.50). Geometric smoothness/roughness parameters Ra, Rq, and Rz showed particularly high intercorrelations (ρ > 0.95), as all represent surface height deviations similarly. Among these, Ra—the arithmetic mean roughness—is widely recognized and thus selected as the representative parameter 20 . Similarly, friction parameters SF and KF were highly correlated (ρ = 0.99). KF was selected as the representative parameter due to its closer relationship with tactile perception, as it directly reflects resistance experienced during finger movement at constant speed. During tactile exploration, participants primarily perceive dynamic frictional properties, making KF more relevant than SF 13 . Following this feature selection, nine properties were retained: KF, Ra, RSm, Rsk, Rku, Rdq, SH, ML, and EB. These features exhibit acceptable VIF levels and comprehensively represent essential mechanical properties for robust tactile perception prediction modeling (Fig. 2 ). Regression Model Construction and Validation Six regression models were utilized to analyze the relationship between fiber mechanical properties and smoothness/roughness perception: linear regression, Gaussian process regression, support vector regression, random forest, gradient boosting, and neural network regression. Model performance and generalization capability were evaluated through 5-fold cross-validation 5 . The dataset (N = 50) was randomly partitioned into five subsets, with four subsets (N = 40) used for training and the remaining subset (N = 10) used for validation. This procedure was repeated five times, allowing each subset to serve once as a validation set, thereby utilizing all data evenly in the evaluation. The predictive performance was assessed using the coefficient of determination (R²), which indicates how effectively a model explains variance in the data, ranging from 0 to 1, with values closer to 1 representing higher accuracy. The R² was calculated as follows: $$\:\begin{array}{c}{\text{R}}^{2}\:=\:1\:-\:\frac{SSR}{SST}\#\left(2\right)\end{array}$$ where SST (Sum of Squares Total) and SSR (Sum of Squares Residual) are defined as: $$\:\begin{array}{c}SST\:={\sum\:({y}_{i}\:-\:\stackrel{-}{y})}^{2}\#\left(3\right)\end{array}$$ $$\:\begin{array}{c}SSR\:={\sum\:({\widehat{y}}_{i}\:-\:{y}_{i})}^{2}\#\left(4\right)\end{array}$$ Here, \(\:{y}_{i}\) represents the actual values, \(\:\stackrel{-}{y}\) is the mean of the actual values, and \(\:{\widehat{y}}_{i}\) represents the predicted values. In this study, model performance was evaluated using two methods: first, a single evaluation by training and testing on the entire dataset (non-cross-validation), and second, using 5-fold cross-validation. This dual approach detects potential overestimation of performance when models are evaluated on training data alone, providing a more accurate estimate of generalization capability. The single evaluation method aligns with previous tactile perception research, facilitating direct comparisons, but it does not reliably reflect performance on unseen samples—an essential factor in practical applications such as material design and quality control. Therefore, we report both single evaluation results for consistency with prior work and cross-validation results to better represent expected performance on novel samples. Comparing these two evaluation methods also helps identify potential overestimation of model performance and enhances understanding of model generalizability. To optimize predictive performance, a feature selection process was implemented. After identifying the nine features with acceptable VIF values, we systematically evaluated each regression model across all possible feature combinations using 5-fold cross-validation. This procedure identified the optimal set of features that maximized predictive performance for each regression technique. Additionally, we analyzed the frequency of feature selection across all optimal models to identify mechanical properties consistently recognized as important predictors of smoothness/roughness perception. Results Correlation between Mechanical Properties and Smoothness/Roughness Perception Figure 3 presents scatter plots illustrating correlations between each of the nine normalized mechanical properties and smoothness/roughness perception. Among the measured properties, KF demonstrated the strongest negative correlation (ρ = -0.67). Ra and RSm showed moderate positive correlations (ρ = 0.44 and ρ = 0.42, respectively), while ML exhibited a moderate negative correlation (ρ = -0.41). Rdq displayed a relatively weak positive correlation (ρ = 0.31), and EB showed a weak negative correlation (ρ = -0.26). Rku demonstrated a weak correlation (ρ = 0.17). Notably, SH (ρ = -0.09) and Rsk (ρ = 0.01) showed negligible correlation with smoothness/roughness perception. Regression Analysis Results The predictive performance of six different regression models for smoothness/roughness perception was evaluated by comparing non-cross-validation and 5-fold cross-validation results. In the non-cross-validation, the gradient boosting and neural network models demonstrated the highest explanatory power (both R² = 1.00), followed by the Gaussian process regression (R² = 0.94). Random forest (R² = 0.81), linear regression (R² = 0.68), and support vector regression (R² = 0.66) showed comparatively lower performance levels (Fig. 4 ). However, the 5-fold cross-validation results revealed substantially different performance patterns (Fig. 5 ). Both the gradient boosting and neural network models, which had perfect performance in the non-cross-validation evaluation, showed notable performance drops to R² = 0.47 and R² = 0.14, respectively. Instead, Gaussian process regression demonstrated the highest performance (R² = 0.65), followed by support vector regression (R² = 0.55), and linear regression (R² = 0.53). Random forest (R² = 0.48) and gradient boosting (R² = 0.47) models exhibited moderate performance. These performance differences between non-cross-validation and cross-validation results indicate potential performance overestimation in some models, particularly in gradient boosting, neural network models, and random forest. Analyzing the performance differences in detail, the linear regression (R² = 0.68 → 0.53) and support vector regression models (R² = 0.66 → 0.55) displayed relatively minor differences non-cross-validation and cross-validation, suggesting stable predictive capabilities. Gaussian process regression consistently maintained strong performance relative to other models, both non-cross-validation and cross-validation (R² = 0.94 → 0.65), despite exhibiting a notable performance drop between these conditions. To improve the prediction performance of the regression models, a feature selection approach was implemented. Performance was systematically evaluated for all possible combinations of the nine features selected after the multicollinearity analysis to determine the optimal set for each model. The Gaussian process regression model achieved the highest predictive performance (R² = 0.71) with six features (KF, Ra, RSm, Rsk, Rku, ML). Linear regression and support vector regression also exhibited strong predictive performance (R² = 0.63 and R² = 0.62, respectively) using the same six-feature set. The random forest regression yielded an R² of 0.59 with five features (KF, RSm, Rsk, Rku, Rdq), while the gradient boosting model achieved R² = 0.51 using a different set of five features (KF, Ra, Rsk, Rku, ML). Neural network regression showed the lowest optimal performance (R² = 0.45) using only three features (KF, Rku, SH) (Fig. 6 ). To provide a comprehensive comparison of model performance, Table 1 presents the results of three evaluation scenarios: no cross-validation, 5-fold cross-validation without feature selection, and 5-fold cross-validation with optimal feature selection. Table 1 Comparison of regression model performance with and without cross-validation and feature optimization. Regression Model Linear Gaussian Process Support Vector Random Forest Gradient Boosting Neural Network Regression Strategy Non-CV 0.68 0.94 0.66 0.81 1.00 1.00 5-fold CV with all features 0.53 0.65 0.55 0.48 0.47 0.14 5-fold CV with optimal features 0.63 0.71 0.62 0.59 0.51 0.45 Analysis of the frequency at which each mechanical property appeared in optimal feature combinations across the six regression models revealed that KF and Rku were the most frequently selected, each appearing in all six models. Rsk was selected five times, while Ra, RSm, and ML were each included four times. Conversely, SH and Rdq were selected only once each, and notably, EB was not included in any optimal feature combinations (Fig. 7 ). Discussion This study developed a predictive model for human smoothness/roughness perception by integrating multiple measurable mechanical properties with subjective evaluations. Unlike earlier approaches that relied on limited features or artificial samples, our model emphasizes both dimensional richness and practical realism. By leveraging a diverse set of 50 commercially available synthetic fibers, we evaluated tactile perception under conditions that closely resemble real-world material interactions. Although fibers served as our experimental platform, the methodology can be broadly applied to a wide range of tactile materials and use cases. The present research offers three primary methodological contributions. First, by systematically selecting nine representative mechanical properties from friction, surface geometry, hardness, and tensile strength through multicollinearity analysis, the model effectively represented diverse factors influencing tactile perception. This multidimensional approach significantly improved both predictive accuracy and interpretability compared to earlier studies that relied on fewer properties 6 , 27 . Second, the use of 50 commercially available synthetic fibers, including polyester, nylon, and spandex blends, provided enhanced material diversity and realism compared to prior studies using artificially constructed stimuli 11 , 12 . Third, implementing a rigorous 5-fold cross-validation procedure ensured robust evaluation of model generalizability, addressing the inflated performance reported by studies lacking independent validation 11 , 15 . Cross-validation results revealed substantial performance discrepancies across regression models, highlighting the critical risk of performance overestimation in modeling tactile perception. Specifically, gradient boosting and neural network models showed perfect performance in non-cross-validation (R² = 1.00) but showed notable declines to R² = 0.47 and R² = 0.14, respectively, after cross-validation. These findings demonstrate that previous research without proper validation may have significantly overestimated prediction accuracy 23 , 26 . Conversely, Gaussian process regression consistently maintained strong performance (R² = 0.71), suggesting superior reliability for practical applications. Linear regression and support vector regression also showed stable predictive capacity across validations, further underscoring the value of cross-validation in ensuring model generalization. Correlation analysis and systematic feature selection revealed complex interactions among mechanical properties underlying tactile smoothness/roughness perception. Kinetic friction (KF) exhibited the strongest individual correlation (ρ = -0.67, p < 0.05), aligning with previous research emphasizing the importance of friction in tactile evaluations 6 , 17 . Additionally, surface geometry (Ra and RSm) and tensile strength (ML) demonstrated moderate but meaningful correlations (ρ ≈ ±0.4, p < 0.05). Notably, although Rku individually exhibited a relatively weak correlation (ρ = 0.17, p = 0.24), its universal selection across all six models highlights its critical importance within multivariate contexts. Similarly, skewness (Rsk), despite showing negligible individual correlation (ρ = 0.01, p = 0.95), was selected in five models. These results indicate that even when individual surface features appear statistically insignificant on their own, they can meaningfully contribute to perceptual prediction when considered jointly with other properties. This suggests that subtle aspects of surface shape and profile distribution, such as asymmetry and peakedness, play a significant role in the integration of tactile information. These findings reinforce the perspective that tactile perception arises from the interaction of multiple mechanical characteristics, rather than being driven by any single factor alone 8 . Several limitations of this study should be noted. First, the study exclusively investigated synthetic fibers, which may limit applicability to natural fibers with differing structural properties. Second, while participant sample size (n = 37) was larger than that of previous tactile studies, broader demographic diversity could further enhance model generalizability. Third, the standardized evaluation protocol (horizontal finger movement using only the distal phalanx) may not fully represent diverse tactile exploration behaviors typical in everyday material interactions. Additionally, individual variations in tactile sensitivity were not explicitly modeled, potentially influencing perceptual assessments. Future research could address these limitations by expanding the sample set to include natural fibers, increasing participant demographic diversity, and incorporating a broader range of tactile exploration behaviors and individual differences in tactile sensitivity. Furthermore, designing custom fiber samples to systematically manipulate the identified key mechanical properties (KF, Ra, RSm, Rsk, Rku) would enable deeper empirical validation and causal insights. Moreover, the proposed modeling framework could be expanded to encompass additional tactile attributes, such as softness, hardness, and thermal perception. It could also be adapted for applications involving wearable haptics, virtual materials, and human–robot interfaces. Therefore, our approach contributes toward the broader goal of developing generalizable tactile models applicable across diverse industries and material categories. Declarations Acknowledgements: This work was partly supported by the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. RS-2023-00302489, 40%) and by Institute of Information & Communications Technology Planning & Evaluation(IITP)-ITRC(Information Technology Research Center) grant funded by the Korea government (MSIT) (IITP-2025-RS-2023-00258971, 30%) and by Institute of Information & communications Technology Planning & Evaluation (IITP) grant funded by the Korea government(MSIT) (No.RS-2025-02263277, 30%). Funding : This work was partly supported by the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. RS-2023-00302489, 40%) and by Institute of Information & Communications Technology Planning & Evaluation(IITP)-ITRC(Information Technology Research Center) grant funded by the Korea government (MSIT) (IITP-2025-RS-2023-00258971, 30%) and by Institute of Information & communications Technology Planning & Evaluation (IITP) grant funded by the Korea government(MSIT) (No.RS-2025-02263277, 30%). Author contributions declaration : Hyung-Tak Lee, Jun-Yeop Kim, and Keungyonh Bak contributed equally to this work and share first authorship. They were primarily responsible for designing the experiment, analyzing the data, developing the regression models, and writing the initial draft of the manuscript. Kiun Kim contributed to the implementation of data collection and assisted in generating figures. Sungwoo Chun and Han-Jeong Hwang , as corresponding authors, supervised the entire research process and provided essential guidance throughout the study. All authors had full access to and responsibility for all aspects of the study. Data availability statement : This study is based on data collected independently by the authors, specifically questionnaire responses from 37 participants and mechanical property data extracted from 50 synthetic fiber samples. The data are not publicly available but can be obtained upon reasonable request from the corresponding author, Han-Jeong Hwang ( [email protected] ), or the first author, Hyung-Tak Lee ( [email protected] ). Conflict of interest: No conflicts of interest. Consent to participate and publish: The study was approved by the Institutional Review Board of Korea University (IRB-2024-0163). Before the experiment, participants were informed about the study objectives, procedures, and necessary precautions, after which they provided informed consent. Appropriate compensation was provided upon completion of the experiment. References See, A. R., Choco, J. A. G. & Chandramohan, K. Touch, Texture and Haptic Feedback: A Review on How We Feel the World around Us. Appl. Sci. 12 , 4686 (2022). Di Stefano, N. & Spence, C. Roughness perception: A multisensory/crossmodal perspective. Atten. Percept. Psychophys. 84 , 2087–2114 (2022). Lipomi, D. J., Dhong, C., Carpenter, C. W., Root, N. B. & Ramachandran, V. S. Organic Haptics: Intersection of Materials Chemistry and Tactile Perception. Adv. Funct. Mater. 30 , 1906850 (2020). Wang, Y. et al. Superhydrophobic wearable sensor: fabrication, application, and perspective. Discov. Nano 19 , 176 (2024). Taudt, C., Leyens, C., Hartmann, P. & Molotnikov, A. Darkfield-scattering surface analysis in powder-based AM-processes: analysis of a multiwavelength approach. in Laser 3D Manufacturing XII (eds. Helvajian, H., Gu, B. & Chen, H.) 16 (SPIE, San Francisco, United States, 2025). doi:10.1117/12.3043198. Tang, W. et al. From finger friction to brain activation: Tactile perception of the roughness of gratings. J. Adv. Res. 21 , 129–139 (2020). Park, J., Yoo, H.-S., Hong, K. H. & Kim, E. Knitted fabric properties influencing coolness to the touch and the relationship between subjective and objective coolness measurements. Text. Res. J. 88 , 1931–1942 (2018). Degraen, D., Piovarči, M., Bickel, B. & Krüger, A. Capturing Tactile Properties of Real Surfaces for Haptic Reproduction. in The 34th Annual ACM Symposium on User Interface Software and Technology 954–971 (ACM, Virtual Event USA, 2021). doi:10.1145/3472749.3474798. Sousa, E. et al. Tactile and visual perception of plastic textures for car interiors: Psychophysical and affective dimensions. Int. J. Ind. Ergon. 92 , 103369 (2022). Sutu, A., Meftah, E.-M. & Chapman, C. E. Physical determinants of the shape of the psychophysical curve relating tactile roughness to raised-dot spacing: implications for neuronal coding of roughness. J. Neurophysiol. 109 , 1403–1415 (2013). Chiba, T., Kuroda, S. & Yamaguchi, M. Modeling the relationship between tactile sensation and physical properties of synthetic leather. J. Ind. Text. 50 , 346–363 (2020). Wu, J., Li, N., Liu, W., Song, G. & Zhang, J. Experimental Study on the Perception Characteristics of Haptic Texture by Multidimensional Scaling. IEEE Trans. Haptics 8 , 410–420 (2015). Lee, H.-T., Cheon, H.-R., Lee, S.-H., Shim, M. & Hwang, H.-J. Publisher Correction: Risk of data leakage in estimating the diagnostic performance of a deep-learning-based computer-aided system for psychiatric disorders. Sci. Rep. 14 , 2172 (2024). James, G., Witten, D., Hastie, T., Tibshirani, R. & Taylor, J. Linear Model Selection and Regularization. in An Introduction to Statistical Learning 229–288 (Springer International Publishing, Cham, 2023). doi:10.1007/978-3-031-38747-0_6. Skedung, L., Harris, K. L., Collier, E. S. & Rutland, M. W. The finishing touches: the role of friction and roughness in haptic perception of surface coatings. Exp. Brain Res. 238 , 1511–1524 (2020). In Hwan Sul, Kyung Hwa Hong, Shim, H., & Tae Jin Kang. Surface Roughness Measurement of Nonwovens Using Three-dimensional Profile Data. Text. Res. J. 76 , 828–834 (2006). Fehlberg, M., Monfort, E., Saikumar, S., Drewing, K. & Bennewitz, R. Perceptual Constancy in the Speed Dependence of Friction During Active Tactile Exploration. IEEE Trans. Haptics 17 , 957–963 (2024). Yeo, J. C. et al. Wearable Mechanotransduced Tactile Sensor for Haptic Perception. Adv. Mater. Technol. 2 , 1700006 (2017). Tadesse, M. G., Chen, Y., Wang, L., Nierstrasz, V. & Loghin, C. Tactile Comfort Prediction of Functional Fabrics from Instrumental Data Using Intelligence Systems. Fibers Polym. 20 , 199–209 (2019). Skedung, L. et al. Feeling Small: Exploring the Tactile Perception Limits. Sci. Rep. 3 , 2617 (2013). Nuriyev, M. N. & Jafarova, A. M. Development of a method for the combined control of the hardness of winding textile package. EUREKA Phys. Eng. 74–84 (2022) doi:10.21303/2461-4262.2022.002237. Turl, L. H. The Measurement of Tearing Strength of Textile Fabrics. Text. Res. J. 26 , 169–176 (1956). Hu, Y., Hu, J., Zhao, Q., Ding, X. & Yang, X. Relationship between tactual roughness judgment and surface morphology of fabric by fingertip touching method. Fibers Polym. 14 , 1024–1031 (2013). Krstajic, D., Buturovic, L. J., Leahy, D. E. & Thomas, S. Cross-validation pitfalls when selecting and assessing regression and classification models. J. Cheminformatics 6 , 10 (2014). Shrestha, N. Detecting Multicollinearity in Regression Analysis. Am. J. Appl. Math. Stat. 8 , 39–42 (2020). Simunovic, G. et al. Surface roughness assessing based on digital image features. Adv. Prod. Eng. Manag. 11 , 93–104 (2016). Karim, A. K. M. R., Prativa, S. & Likova, L. Perception and appreciation of tactile objects: The role of visual experience and texture parameters (JPI-first). J. Percept. Imaging 34 , (2022). Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 27 Nov, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 18 Sep, 2025 Reviews received at journal 17 Sep, 2025 Reviewers agreed at journal 14 Sep, 2025 Reviews received at journal 14 Sep, 2025 Reviewers agreed at journal 09 Sep, 2025 Reviews received at journal 08 Sep, 2025 Reviewers agreed at journal 07 Sep, 2025 Reviewers agreed at journal 06 Sep, 2025 Reviewers agreed at journal 06 Sep, 2025 Reviewers invited by journal 06 Sep, 2025 Editor assigned by journal 11 Aug, 2025 Submission checks completed at journal 07 Jul, 2025 First submitted to journal 07 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7046348","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":512327060,"identity":"52a30ebd-42cf-4b3f-b6d7-b9fd6ddbfc53","order_by":0,"name":"Hyung-Tak Lee","email":"","orcid":"","institution":"Korea University","correspondingAuthor":false,"prefix":"","firstName":"Hyung-Tak","middleName":"","lastName":"Lee","suffix":""},{"id":512327061,"identity":"bbb61e73-117d-4fef-aabe-f771841e24ef","order_by":1,"name":"Jun-Yeop Kim","email":"","orcid":"","institution":"Korea University","correspondingAuthor":false,"prefix":"","firstName":"Jun-Yeop","middleName":"","lastName":"Kim","suffix":""},{"id":512327062,"identity":"9df4ea23-ca65-487a-a368-c9c1c4a8e450","order_by":2,"name":"Keungyonh Bak","email":"","orcid":"","institution":"Korea University","correspondingAuthor":false,"prefix":"","firstName":"Keungyonh","middleName":"","lastName":"Bak","suffix":""},{"id":512327063,"identity":"fa7362fa-5a74-4600-88f0-00f9bca90edd","order_by":3,"name":"Kiun KIM","email":"","orcid":"","institution":"Korea University","correspondingAuthor":false,"prefix":"","firstName":"Kiun","middleName":"","lastName":"KIM","suffix":""},{"id":512327064,"identity":"10b7af5e-eec6-4273-a645-449e7a80cf2f","order_by":4,"name":"Sungwoo Chun","email":"","orcid":"","institution":"Korea University","correspondingAuthor":false,"prefix":"","firstName":"Sungwoo","middleName":"","lastName":"Chun","suffix":""},{"id":512327065,"identity":"08bee214-4884-401a-b51b-d188e15d239a","order_by":5,"name":"Han-Jeong Hwang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3UlEQVRIiWNgGAWjYBACAxDB2MAmJwEX4iFSizHJWhgSZxCtxVwi+dnDrzv40me29z588IPBTp6B5+wDvFosZ6SZG8ueYcudzXPc2LCHIdmwgbfdAL/DbieYSUu2seXOk0hjk+BhYE5g4Gcj4Jfb6d9AWtLlJNLYf/5hqCdGS46Z5Mc2tgRpoC3MPAyHExh42whouf+mTJqxjc1wZs8xZmkZg+OGbTzHCGg5c3yb5M+2Y/ISx9sYP76pqJbn50nDrwUEgO6BGQwMKwI+gQDGHww1xKgbBaNgFIyCkQoANz478dTWXmQAAAAASUVORK5CYII=","orcid":"","institution":"Korea University","correspondingAuthor":true,"prefix":"","firstName":"Han-Jeong","middleName":"","lastName":"Hwang","suffix":""}],"badges":[],"createdAt":"2025-07-04 11:38:03","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7046348/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7046348/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-26294-5","type":"published","date":"2025-11-27T15:57:11+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":91195627,"identity":"6c917362-f60d-493c-b149-5a6bc0e740c6","added_by":"auto","created_at":"2025-09-12 14:58:41","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":139283,"visible":true,"origin":"","legend":"\u003cp\u003eDistribution of samples across smoothness/roughness perception levels. The boxplot illustrates the number of participant responses for each perception level. Lower values indicate smoother textures, while higher values indicate rougher textures.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-7046348/v1/ae4bce18807cde265c854795.png"},{"id":91196800,"identity":"e5bfc385-f8c5-4d87-aad7-abbb598ffa75","added_by":"auto","created_at":"2025-09-12 15:06:41","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":39393,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelation matrix of mechanical properties illustrating multicollinearity. Among highly correlated parameters, KF was selected over SF, and Ra was chosen among Ra, Rq, and Rz as representative features.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-7046348/v1/ad84619701b6461c346d486d.png"},{"id":91199694,"identity":"e33cff1c-0009-4018-91d2-66f97298ec1b","added_by":"auto","created_at":"2025-09-12 15:22:41","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":547900,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelation between each of the nine normalized mechanical properties and smoothness/roughness perception.\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-7046348/v1/b39ad849ee5996e17bcff585.png"},{"id":91195629,"identity":"d2f3ded7-3cf3-45d2-9165-ebfd0ad4396d","added_by":"auto","created_at":"2025-09-12 14:58:41","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":342978,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of regression model performance for predicting smoothness/roughness perception using non-cross-validated results (entire dataset for training and testing).\u003c/p\u003e","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-7046348/v1/e7be3fdce3ff47254d6b4fa9.png"},{"id":91196803,"identity":"abe1ea51-ae29-4450-aff9-531c99054d03","added_by":"auto","created_at":"2025-09-12 15:06:41","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":413394,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of regression model performance for predicting smoothness/roughness perception using cross-validated results.\u003c/p\u003e","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-7046348/v1/de3bcdac8e53b4417edf1d26.png"},{"id":91195635,"identity":"238befc0-92cd-4f82-89c4-4c62967cf403","added_by":"auto","created_at":"2025-09-12 14:58:41","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":415746,"visible":true,"origin":"","legend":"\u003cp\u003eOptimal performance of regression models after feature selection using 5-fold cross-validation.\u003c/p\u003e","description":"","filename":"Fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-7046348/v1/ff400b075252f998465608a3.png"},{"id":91199696,"identity":"aca73a49-baa7-42fc-8cba-db3e87b6493c","added_by":"auto","created_at":"2025-09-12 15:22:41","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":33422,"visible":true,"origin":"","legend":"\u003cp\u003eFrequency of mechanical properties selected in optimal feature combinations across the six regression models.\u003c/p\u003e","description":"","filename":"Fig7.png","url":"https://assets-eu.researchsquare.com/files/rs-7046348/v1/f4e616c55d88eb9e9d5787db.png"},{"id":97178759,"identity":"64b4c1ea-fc42-4a26-a3e5-64aceebbf5e4","added_by":"auto","created_at":"2025-12-01 16:13:22","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2370485,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7046348/v1/36e05208-37a5-4e46-b98b-9b956efccd2b.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Predicting Human Tactile Smoothness/Roughness Perception from Multidimensional Mechanical Properties of Synthetic Fibers Using Machine Learning","fulltext":[{"header":"Introduction","content":"\u003cp\u003eTactile perception refers to the subjective sensation formed by integrating information received through receptors distributed across the skin. This perception includes tactile attributes such as smoothness/roughness, softness/hardness, coldness/warmth, and springiness/stiffness. These tactile attributes play essential roles not only in evaluating material comfort, but also in shaping user experience across a wide range of domains, including consumer products, virtual reality, robotics, and rehabilitation \u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e–\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Smoothness/roughness is particularly influential in how users judge surface quality and usability in both daily interactions and specialized applications. Therefore, developing reliable methods for evaluating smoothness/roughness is crucial for quality control, material development, and consumer satisfaction across industries.\u003c/p\u003e\u003cp\u003eGiven this importance, researchers have proposed various quantitative methods to evaluate smoothness/roughness based on surface mechanical properties, ranging from surface friction measurements to advanced analyses of nanoscale microstructures. Recent studies have achieved remarkable spatial resolutions (10 nm–1 µm) using sophisticated measurement techniques for surface smoothness/roughness measurement \u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. However, these quantitative mechanical measurements alone cannot fully capture human perceptions of smoothness/roughness \u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e because such perception involves subjective interpretations influenced by multisensory integration, context, and individual variability. Consequently, integrating quantitative mechanical measurements with subjective evaluations of smoothness/roughness is increasingly recognized as essential.\u003c/p\u003e\u003cp\u003eEarly studies primarily explored relationships between perceived smoothness or roughness and individual mechanical properties, such as surface geometry or frictional force. More recent research emphasizes that tactile perception arises from the complex interactions among multiple mechanical factors, including surface topography, height variation, spacing patterns, and material composition \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Multiparameter modeling approaches have consistently outperformed single-feature methods, showing significant improvements in predictive performance (R²) of approximately 0.20 to 0.35 \u003csup\u003e8,9\u003c/sup\u003e. These findings highlight the necessity of multidimensional approaches that comprehensively integrate geometric properties (e.g., surface height and spacing), contact-related properties (e.g., friction and hardness), and intrinsic material properties for accurate tactile perception modeling.\u003c/p\u003e\u003cp\u003eWhile multidimensional approaches are essential, achieving a balance between experimental control and practical relevance is equally important. Most previous studies used artificially manufactured samples via 3D printing or lithography to ensure precise control. Although these studies reported high predictive accuracy (classification accuracies and regression R² values ranging from approximately 0.75 to 0.89), their generalizability to real-world materials, influenced by multiple uncontrolled factors, remains limited \u003csup\u003e\u003cspan additionalcitationids=\"CR11\" citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e–\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. Artificial samples typically simplify complex real-world textures by varying only a few parameters, insufficiently capturing the intricate interplay among mechanical factors. To overcome this limitation, it is necessary to investigate perception using realistic materials that naturally embody diverse mechanical interactions.\u003c/p\u003e\u003cp\u003eAs research increasingly emphasizes linking mechanical properties to human tactile perception using real-world samples, developing predictive models with practical generalizability has become crucial. Achieving this involves multidimensional feature integration coupled with rigorous validation. Cross-validation, particularly, ensures models accurately predict tactile responses for unseen materials. Recent studies by Lee et al. and James et al. illustrated that neglecting proper cross-validation procedures leads to inflated accuracy, dramatically reducing predictive reliability on new data \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Although previous tactile studies reported high performance (R² values of 0.72–0.92), many evaluated their models using the same datasets used for training, limiting their applicability \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. To ensure reliable and transferable outcomes, predictive tactile models must incorporate robust cross-validation procedures.\u003c/p\u003e\u003cp\u003eThis study aims to develop a predictive model for human smoothness/roughness perception based on multiple mechanical properties of fiber materials, making three significant contributions to tactile perception research. First, we systematically selected nine representative mechanical features from an initial set of twelve through multicollinearity analysis, enabling comprehensive yet non-redundant characterization of fiber surfaces. This approach represents a methodological improvement compared to earlier studies that typically considered fewer parameters without systematic feature selection. Second, we used 50 diverse synthetic fibers commonly encountered in daily life, including polyester, spandex, nylon, and their composites. This sample size significantly expands upon the relatively small number of samples (typically 10–20) used in earlier research. Third, we rigorously evaluated model performance through five-fold cross-validation, addressing previous limitations of studies reporting high predictive accuracy without independent validation. Utilizing these selected features, we developed and evaluated six regression models, determining the most effective approach for predicting smoothness/roughness perception. Our approach provides a robust and transferable framework for modeling tactile perception, with potential applications in material evaluation, product development, and haptic technology design. Furthermore, the established framework has the potential to be extended to other tactile dimensions such as softness, hardness, and thermal sensations.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cb\u003eMaterial Preparation\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn this study, 50 synthetic fibers commonly used in practical applications were selected for experimental analysis. The fiber samples were classified into single-component and composite groups. The single-component group consisted of 30 samples of 100% polyester and 2 samples of 100% nylon. The composite group included 18 blended samples: polyester-based (9 polyester/spandex, 2 polyester/polyurethane combinations) and nylon-based blends (5 nylon/spandex, 2 nylon/polyurethane combinations). These fiber types were chosen to reflect materials frequently encountered in textile and apparel products, enabling the development of a practically relevant model for predicting tactile smoothness/roughness perception.\u003c/p\u003e\u003cp\u003ePolyester, representing approximately 70% of the global fiber market, was primarily selected due to its significant market share, durability, cost-effectiveness, and widespread industrial and consumer applications \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Nylon and synthetic blends were also included because of their common use in functional and everyday clothing, offering diverse tactile and mechanical characteristics such as variations in friction, elasticity, and surface texture. The inclusion of blended materials particularly enhances the model's applicability, reflecting the complexity of real-world tactile experiences.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eMechanical Property Measurement\u003c/b\u003e\u003c/p\u003e\u003cp\u003eHuman tactile smoothness/roughness perception arises from the integration of various stimuli detected by mechanoreceptors in the skin during surface contact. To accurately model this complex perceptual process, we measured 12 mechanical properties across four categories that comprehensively characterize fiber surfaces: First, two frictional properties of the samples were assessed using standardized testing methods. Although these tests may not fully replicate finger-surface interactions, they allow objective comparisons among materials and have been validated as reliable predictors of tactile perception in numerous studies \u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. Second, geometric roughness properties were evaluated through seven parameters defined in the ISO 4287 standard. Third, Shore hardness was measured to account for the influence of material hardness on tactile sensation. Hardness has been demonstrated by Yeo et al. (2017) to significantly impact tactile feedback during contact \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. Lastly, two tensile properties, including strength and extensibility, were examined to reflect material deformation under tactile interaction forces. These mechanical attributes significantly contribute to tactile experiences in textile perception \u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. The twelve selected properties collectively represent diverse aspects of tactile interactions between human skin and fiber surfaces. This multidimensional approach effectively captures the complex microstructure characteristics that cannot be adequately represented by a single roughness parameter \u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. Each property was measured 10 times per sample, and mean values were calculated to ensure measurement reliability.\u003c/p\u003e\u003cp\u003e\u003cb\u003eFriction Force Measurement\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe frictional characteristics of the fiber samples were quantified by measuring the static (SF) and kinetic friction (KF) coefficients using a friction coefficient tester (QMESYS, Korea) \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Measurement conditions included a load cell of 1.0 kgf, movement speed of 200.0 mm/min, and friction weight of 200.0 g. Friction force was measured by securing each fiber sample to the friction element. The SF coefficient was recorded at the initiation of movement, while the KF coefficient was measured during constant velocity motion.\u003c/p\u003e\u003cp\u003e\u003cb\u003eGeometric Smoothness/Roughness Measurement\u003c/b\u003e\u003c/p\u003e\u003cp\u003eGeometric smoothness/roughness of the fiber samples was measured using an Alpha-Step IQ surface profiler (KLA Tencor, USA) \u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. Measurement parameters included a sampling rate of 200 Hz, scan length of 5.0 mm, and scan speed of 0.1 mm/sec. Seven parameters were extracted from the profiles based on ISO 4287 standard: arithmetic mean roughness (Ra), root mean square roughness (Rq), maximum height of the profile (Rz), mean width of the profile elements (RSm), skewness of the profile (Rsk), kurtosis of the profile (Rku), and root mean square slope (Rdq).\u003c/p\u003e\u003cp\u003e\u003cb\u003eHardness Measurement\u003c/b\u003e\u003c/p\u003e\u003cp\u003eShore hardness (SH) of fiber samples was measured using an HT-6510 hardness testers (REED Instruments, USA) \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. The tester was mounted on a support stand, and SH was measured at multiple points on each fiber sample to obtain representative hardness values.\u003c/p\u003e\u003cp\u003e\u003cb\u003eTensile Strength Measurement\u003c/b\u003e\u003c/p\u003e\u003cp\u003eMaximum load (ML) and elongation at break (EB) of the fiber samples were measured using Instron tensile testing machine (Instron, USA) \u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. Measurements were conducted at a tensile speed of 100.0 mm/min. ML represents the peak force that fibers withstand before breaking, while EB is expressed as the percentage of extended length relative to initial length at rupture, quantitatively evaluating the strength and extensibility of each fiber sample.\u003c/p\u003e\u003cp\u003e\u003cb\u003eTactile Perception Evaluation\u003c/b\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eParticipants\u003c/b\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe experiment was conducted with 37 healthy adult volunteers (23 males and 14 females; mean age: 24.1 ± 2.6 years). Participants had no mental or physical disorders, nor any history of sensory nerve damage. The study protocol was approved by the Institutional Review Board of Korea University (IRB-2024-0163). All processes were conducted in accordance with the relevant guidelines and regulations, including the principles outlined in the Declaration of Helsinki. Before the experiment, participants were informed about the study objectives, procedures, and necessary precautions, after which they provided informed consent. Appropriate compensation was provided upon completion of the experiment.\u003c/p\u003e\u003cp\u003e\u003cb\u003eExperimental Procedure\u003c/b\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe tactile smoothness/roughness perception experiment was conducted using 50 fiber samples. Participants' hands were thoroughly cleaned prior to evaluation to eliminate interference from foreign substances or perspiration. Detailed instructions on the evaluation method were provided, and participants practiced with sample materials until they were comfortable with the procedure. All fiber samples were stored under identical environmental conditions and presented in random order.\u003c/p\u003e\u003cp\u003eTo minimize perceptual variations, tactile perception was assessed using only the distal phalanx of the right index finger \u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. Participants moved their index finger horizontally across each fiber sample, ensuring consistent tactile stimulation and reliable assessments. Participants classified each sample on a 5-level scale, with lower values indicating smoother textures and higher values indicating rougher textures. This 5-point classification scheme is consistent with established practices in tactile research, effectively capturing meaningful differences in human tactile perception with optimal sensitivity and minimal complexity [31–33]. The use of this approach ensured a balanced distribution of samples across perception levels (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), providing adequate statistical power and reducing potential bias during regression analyses, particularly within the cross-validation procedure for assessing predictive model generalizability. The responses from all 37 participants were averaged for each fiber sample to derive representative tactile perception values, which were subsequently used for modeling the correlation between human tactile perception and physical properties.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eData Analysis and Modeling\u003c/b\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eData Preprocessing\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe mechanical properties possess varying units and ranges, complicating their direct comparison in smoothness/roughness perception prediction models. For instance, Ra is measured in micrometers (µm), two friction coefficients (SF and KF) are dimensionless values, and SH uses a scale from 0-100. Such scale discrepancies may cause disproportionate influence of certain features or underestimation of important ones during model training. To resolve this issue, each mechanical property was normalized using standard scores (z-scores) \u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e, calculated as follows:\u003c/p\u003e\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{c}Z\\:=\\:\\frac{\\text{X}\\:-\\:{\\mu\\:}\\:}{{\\sigma\\:}}\\#\\left(1\\right)\\end{array}$$\u003c/div\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003ewhere Z is the standardized value, X represents the original data value, µ is the mean of the respective feature, and σ is the standard deviation of the feature. This normalization transforms all mechanical properties into distributions with a mean of 0 and a standard deviation of 1, facilitating direct feature comparison. Normalized data enhances the accuracy of feature importance assessment and improves the convergence and stability of machine learning algorithms.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eMulticollinearity Analysis\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAddressing multicollinearity among independent variables is essential for accurate and reliable regression modeling. Multicollinearity occurs when independent variables exhibit high correlations, potentially destabilizing regression coefficients and complicating model interpretation. Specifically, including highly correlated variables together makes it difficult to assess their distinct effects, which can negatively impact model performance \u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. To identify multicollinearity issues, Variance Inflation Factor (VIF) analysis was conducted on the initial 12 mechanical properties. A VIF value exceeding 10 generally indicates severe multicollinearity [27]. The analysis identified severe multicollinearity in SF (153.16), KF (157.19), Rz (217.62), Ra (1151.40), and Rq (2075.50), while acceptable VIF values were observed for Rsk (1.57), Rdq (2.53), RSm (2.78), Rku (4.32), SH (1.67), EB (1.44), and ML (1.50).\u003c/p\u003e\u003cp\u003eGeometric smoothness/roughness parameters Ra, Rq, and Rz showed particularly high intercorrelations (ρ \u0026gt; 0.95), as all represent surface height deviations similarly. Among these, Ra—the arithmetic mean roughness—is widely recognized and thus selected as the representative parameter \u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. Similarly, friction parameters SF and KF were highly correlated (ρ = 0.99). KF was selected as the representative parameter due to its closer relationship with tactile perception, as it directly reflects resistance experienced during finger movement at constant speed. During tactile exploration, participants primarily perceive dynamic frictional properties, making KF more relevant than SF \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Following this feature selection, nine properties were retained: KF, Ra, RSm, Rsk, Rku, Rdq, SH, ML, and EB. These features exhibit acceptable VIF levels and comprehensively represent essential mechanical properties for robust tactile perception prediction modeling (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eRegression Model Construction and Validation\u003c/b\u003e\u003c/p\u003e\u003cp\u003eSix regression models were utilized to analyze the relationship between fiber mechanical properties and smoothness/roughness perception: linear regression, Gaussian process regression, support vector regression, random forest, gradient boosting, and neural network regression. Model performance and generalization capability were evaluated through 5-fold cross-validation \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. The dataset (N = 50) was randomly partitioned into five subsets, with four subsets (N = 40) used for training and the remaining subset (N = 10) used for validation. This procedure was repeated five times, allowing each subset to serve once as a validation set, thereby utilizing all data evenly in the evaluation. The predictive performance was assessed using the coefficient of determination (R²), which indicates how effectively a model explains variance in the data, ranging from 0 to 1, with values closer to 1 representing higher accuracy. The R² was calculated as follows:\u003c/p\u003e\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{c}{\\text{R}}^{2}\\:=\\:1\\:-\\:\\frac{SSR}{SST}\\#\\left(2\\right)\\end{array}$$\u003c/div\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003ewhere SST (Sum of Squares Total) and SSR (Sum of Squares Residual) are defined as:\u003c/p\u003e\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{c}SST\\:={\\sum\\:({y}_{i}\\:-\\:\\stackrel{-}{y})}^{2}\\#\\left(3\\right)\\end{array}$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{c}SSR\\:={\\sum\\:({\\widehat{y}}_{i}\\:-\\:{y}_{i})}^{2}\\#\\left(4\\right)\\end{array}$$\u003c/div\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eHere, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{i}\\)\u003c/span\u003e\u003c/span\u003e represents the actual values, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\stackrel{-}{y}\\)\u003c/span\u003e\u003c/span\u003e is the mean of the actual values, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\widehat{y}}_{i}\\)\u003c/span\u003e\u003c/span\u003e represents the predicted values.\u003c/p\u003e\u003cp\u003eIn this study, model performance was evaluated using two methods: first, a single evaluation by training and testing on the entire dataset (non-cross-validation), and second, using 5-fold cross-validation. This dual approach detects potential overestimation of performance when models are evaluated on training data alone, providing a more accurate estimate of generalization capability. The single evaluation method aligns with previous tactile perception research, facilitating direct comparisons, but it does not reliably reflect performance on unseen samples—an essential factor in practical applications such as material design and quality control. Therefore, we report both single evaluation results for consistency with prior work and cross-validation results to better represent expected performance on novel samples. Comparing these two evaluation methods also helps identify potential overestimation of model performance and enhances understanding of model generalizability.\u003c/p\u003e\u003cp\u003eTo optimize predictive performance, a feature selection process was implemented. After identifying the nine features with acceptable VIF values, we systematically evaluated each regression model across all possible feature combinations using 5-fold cross-validation. This procedure identified the optimal set of features that maximized predictive performance for each regression technique. Additionally, we analyzed the frequency of feature selection across all optimal models to identify mechanical properties consistently recognized as important predictors of smoothness/roughness perception.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cb\u003eCorrelation between Mechanical Properties and Smoothness/Roughness Perception\u003c/b\u003e\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e presents scatter plots illustrating correlations between each of the nine normalized mechanical properties and smoothness/roughness perception. Among the measured properties, KF demonstrated the strongest negative correlation (ρ = -0.67). Ra and RSm showed moderate positive correlations (ρ\u0026thinsp;=\u0026thinsp;0.44 and ρ\u0026thinsp;=\u0026thinsp;0.42, respectively), while ML exhibited a moderate negative correlation (ρ = -0.41). Rdq displayed a relatively weak positive correlation (ρ\u0026thinsp;=\u0026thinsp;0.31), and EB showed a weak negative correlation (ρ = -0.26). Rku demonstrated a weak correlation (ρ\u0026thinsp;=\u0026thinsp;0.17). Notably, SH (ρ = -0.09) and Rsk (ρ\u0026thinsp;=\u0026thinsp;0.01) showed negligible correlation with smoothness/roughness perception.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eRegression Analysis Results\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe predictive performance of six different regression models for smoothness/roughness perception was evaluated by comparing non-cross-validation and 5-fold cross-validation results. In the non-cross-validation, the gradient boosting and neural network models demonstrated the highest explanatory power (both R\u0026sup2; = 1.00), followed by the Gaussian process regression (R\u0026sup2; = 0.94). Random forest (R\u0026sup2; = 0.81), linear regression (R\u0026sup2; = 0.68), and support vector regression (R\u0026sup2; = 0.66) showed comparatively lower performance levels (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eHowever, the 5-fold cross-validation results revealed substantially different performance patterns (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Both the gradient boosting and neural network models, which had perfect performance in the non-cross-validation evaluation, showed notable performance drops to R\u0026sup2; = 0.47 and R\u0026sup2; = 0.14, respectively. Instead, Gaussian process regression demonstrated the highest performance (R\u0026sup2; = 0.65), followed by support vector regression (R\u0026sup2; = 0.55), and linear regression (R\u0026sup2; = 0.53). Random forest (R\u0026sup2; = 0.48) and gradient boosting (R\u0026sup2; = 0.47) models exhibited moderate performance.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThese performance differences between non-cross-validation and cross-validation results indicate potential performance overestimation in some models, particularly in gradient boosting, neural network models, and random forest. Analyzing the performance differences in detail, the linear regression (R\u0026sup2; = 0.68 \u003cspan fontcategory=\"NonProportional\" class=\"\" name=\"Emphasis\"\u003e\u0026rarr;\u003c/span\u003e 0.53) and support vector regression models (R\u0026sup2; = 0.66 \u003cspan fontcategory=\"NonProportional\" class=\"\" name=\"Emphasis\"\u003e\u0026rarr;\u003c/span\u003e 0.55) displayed relatively minor differences non-cross-validation and cross-validation, suggesting stable predictive capabilities. Gaussian process regression consistently maintained strong performance relative to other models, both non-cross-validation and cross-validation (R\u0026sup2; = 0.94 \u003cspan fontcategory=\"NonProportional\" class=\"\" name=\"Emphasis\"\u003e\u0026rarr;\u003c/span\u003e 0.65), despite exhibiting a notable performance drop between these conditions.\u003c/p\u003e\u003cp\u003eTo improve the prediction performance of the regression models, a feature selection approach was implemented. Performance was systematically evaluated for all possible combinations of the nine features selected after the multicollinearity analysis to determine the optimal set for each model. The Gaussian process regression model achieved the highest predictive performance (R\u0026sup2; = 0.71) with six features (KF, Ra, RSm, Rsk, Rku, ML). Linear regression and support vector regression also exhibited strong predictive performance (R\u0026sup2; = 0.63 and R\u0026sup2; = 0.62, respectively) using the same six-feature set. The random forest regression yielded an R\u0026sup2; of 0.59 with five features (KF, RSm, Rsk, Rku, Rdq), while the gradient boosting model achieved R\u0026sup2; = 0.51 using a different set of five features (KF, Ra, Rsk, Rku, ML). Neural network regression showed the lowest optimal performance (R\u0026sup2; = 0.45) using only three features (KF, Rku, SH) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). To provide a comprehensive comparison of model performance, Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e presents the results of three evaluation scenarios: no cross-validation, 5-fold cross-validation without feature selection, and 5-fold cross-validation with optimal feature selection.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eComparison of regression model performance with and without cross-validation and feature optimization.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"8\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colspan=\"6\" nameend=\"c8\" namest=\"c3\"\u003e\u003cp\u003eRegression Model\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLinear\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eGaussian\u003c/p\u003e\u003cp\u003eProcess\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSupport\u003c/p\u003e\u003cp\u003eVector\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eRandom\u003c/p\u003e\u003cp\u003eForest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003eGradient\u003c/p\u003e\u003cp\u003eBoosting\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003eNeural\u003c/p\u003e\u003cp\u003eNetwork\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003eRegression\u003c/p\u003e\u003cp\u003eStrategy\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNon-CV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.68\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.94\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.66\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.81\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e1.00\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e5-fold CV with all features\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.53\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.65\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.55\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.48\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.47\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.14\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e5-fold CV with optimal features\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.63\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.71\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.62\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.59\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.51\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.45\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eAnalysis of the frequency at which each mechanical property appeared in optimal feature combinations across the six regression models revealed that KF and Rku were the most frequently selected, each appearing in all six models. Rsk was selected five times, while Ra, RSm, and ML were each included four times. Conversely, SH and Rdq were selected only once each, and notably, EB was not included in any optimal feature combinations (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study developed a predictive model for human smoothness/roughness perception by integrating multiple measurable mechanical properties with subjective evaluations. Unlike earlier approaches that relied on limited features or artificial samples, our model emphasizes both dimensional richness and practical realism. By leveraging a diverse set of 50 commercially available synthetic fibers, we evaluated tactile perception under conditions that closely resemble real-world material interactions. Although fibers served as our experimental platform, the methodology can be broadly applied to a wide range of tactile materials and use cases.\u003c/p\u003e\u003cp\u003eThe present research offers three primary methodological contributions. First, by systematically selecting nine representative mechanical properties from friction, surface geometry, hardness, and tensile strength through multicollinearity analysis, the model effectively represented diverse factors influencing tactile perception. This multidimensional approach significantly improved both predictive accuracy and interpretability compared to earlier studies that relied on fewer properties \u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. Second, the use of 50 commercially available synthetic fibers, including polyester, nylon, and spandex blends, provided enhanced material diversity and realism compared to prior studies using artificially constructed stimuli \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. Third, implementing a rigorous 5-fold cross-validation procedure ensured robust evaluation of model generalizability, addressing the inflated performance reported by studies lacking independent validation \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eCross-validation results revealed substantial performance discrepancies across regression models, highlighting the critical risk of performance overestimation in modeling tactile perception. Specifically, gradient boosting and neural network models showed perfect performance in non-cross-validation (R\u0026sup2; = 1.00) but showed notable declines to R\u0026sup2; = 0.47 and R\u0026sup2; = 0.14, respectively, after cross-validation. These findings demonstrate that previous research without proper validation may have significantly overestimated prediction accuracy \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. Conversely, Gaussian process regression consistently maintained strong performance (R\u0026sup2; = 0.71), suggesting superior reliability for practical applications. Linear regression and support vector regression also showed stable predictive capacity across validations, further underscoring the value of cross-validation in ensuring model generalization.\u003c/p\u003e\u003cp\u003eCorrelation analysis and systematic feature selection revealed complex interactions among mechanical properties underlying tactile smoothness/roughness perception. Kinetic friction (KF) exhibited the strongest individual correlation (ρ = -0.67, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05), aligning with previous research emphasizing the importance of friction in tactile evaluations \u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. Additionally, surface geometry (Ra and RSm) and tensile strength (ML) demonstrated moderate but meaningful correlations (ρ \u0026asymp; \u0026plusmn;0.4, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Notably, although Rku individually exhibited a relatively weak correlation (ρ\u0026thinsp;=\u0026thinsp;0.17, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.24), its universal selection across all six models highlights its critical importance within multivariate contexts. Similarly, skewness (Rsk), despite showing negligible individual correlation (ρ\u0026thinsp;=\u0026thinsp;0.01, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.95), was selected in five models. These results indicate that even when individual surface features appear statistically insignificant on their own, they can meaningfully contribute to perceptual prediction when considered jointly with other properties. This suggests that subtle aspects of surface shape and profile distribution, such as asymmetry and peakedness, play a significant role in the integration of tactile information. These findings reinforce the perspective that tactile perception arises from the interaction of multiple mechanical characteristics, rather than being driven by any single factor alone \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eSeveral limitations of this study should be noted. First, the study exclusively investigated synthetic fibers, which may limit applicability to natural fibers with differing structural properties. Second, while participant sample size (n\u0026thinsp;=\u0026thinsp;37) was larger than that of previous tactile studies, broader demographic diversity could further enhance model generalizability. Third, the standardized evaluation protocol (horizontal finger movement using only the distal phalanx) may not fully represent diverse tactile exploration behaviors typical in everyday material interactions. Additionally, individual variations in tactile sensitivity were not explicitly modeled, potentially influencing perceptual assessments.\u003c/p\u003e\u003cp\u003eFuture research could address these limitations by expanding the sample set to include natural fibers, increasing participant demographic diversity, and incorporating a broader range of tactile exploration behaviors and individual differences in tactile sensitivity. Furthermore, designing custom fiber samples to systematically manipulate the identified key mechanical properties (KF, Ra, RSm, Rsk, Rku) would enable deeper empirical validation and causal insights. Moreover, the proposed modeling framework could be expanded to encompass additional tactile attributes, such as softness, hardness, and thermal perception. It could also be adapted for applications involving wearable haptics, virtual materials, and human\u0026ndash;robot interfaces. Therefore, our approach contributes toward the broader goal of developing generalizable tactile models applicable across diverse industries and material categories.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements:\u0026nbsp;\u003c/strong\u003eThis work was partly supported by the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. RS-2023-00302489, 40%) and by Institute of Information \u0026amp; Communications Technology Planning \u0026amp; Evaluation(IITP)-ITRC(Information Technology Research Center) grant funded by the Korea government (MSIT) (IITP-2025-RS-2023-00258971, 30%) and by Institute of Information \u0026amp; communications Technology Planning \u0026amp; Evaluation (IITP) grant funded by the Korea government(MSIT) (No.RS-2025-02263277, 30%).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e: This work was partly supported by the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. RS-2023-00302489, 40%) and by Institute of Information \u0026amp; Communications Technology Planning \u0026amp; Evaluation(IITP)-ITRC(Information Technology Research Center) grant funded by the Korea government (MSIT) (IITP-2025-RS-2023-00258971, 30%) and by Institute of Information \u0026amp; communications Technology Planning \u0026amp; Evaluation (IITP) grant funded by the Korea government(MSIT) (No.RS-2025-02263277, 30%).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions declaration\u003c/strong\u003e: \u003cstrong\u003e\u003cem\u003eHyung-Tak Lee, Jun-Yeop Kim, and Keungyonh Bak\u003c/em\u003e\u003c/strong\u003e contributed equally to this work and share first authorship. They were primarily responsible for designing the experiment, analyzing the data, developing the regression models, and writing the initial draft of the manuscript. \u003cstrong\u003e\u003cem\u003eKiun Kim\u003c/em\u003e\u003c/strong\u003e contributed to the implementation of data collection and assisted in generating figures. \u003cstrong\u003e\u003cem\u003eSungwoo Chun and Han-Jeong Hwang\u003c/em\u003e\u003c/strong\u003e, as corresponding authors, supervised the entire research process and provided essential guidance throughout the study. All authors had full access to and responsibility for all aspects of the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability statement\u003c/strong\u003e: This study is based on data collected independently by the authors, specifically questionnaire responses from 37 participants and mechanical property data extracted from 50 synthetic fiber samples. The data are not publicly available but can be obtained upon reasonable request from the corresponding author, Han-Jeong Hwang ([email protected]), or the first author, Hyung-Tak Lee ([email protected]).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of interest:\u0026nbsp;\u003c/strong\u003eNo conflicts of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to participate and publish:\u003c/strong\u003e The study was approved by the Institutional Review Board of Korea University (IRB-2024-0163). Before the experiment, participants were informed about the study objectives, procedures, and necessary precautions, after which they provided informed consent. Appropriate compensation was provided upon completion of the experiment.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eSee, A. R., Choco, J. A. G. \u0026amp; Chandramohan, K. Touch, Texture and Haptic Feedback: A Review on How We Feel the World around Us. \u003cem\u003eAppl. Sci.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 4686 (2022).\u003c/li\u003e\n\u003cli\u003eDi Stefano, N. \u0026amp; Spence, C. Roughness perception: A multisensory/crossmodal perspective. \u003cem\u003eAtten. Percept. Psychophys.\u003c/em\u003e \u003cstrong\u003e84\u003c/strong\u003e, 2087\u0026ndash;2114 (2022).\u003c/li\u003e\n\u003cli\u003eLipomi, D. J., Dhong, C., Carpenter, C. W., Root, N. B. \u0026amp; Ramachandran, V. S. Organic Haptics: Intersection of Materials Chemistry and Tactile Perception. \u003cem\u003eAdv. Funct. Mater.\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 1906850 (2020).\u003c/li\u003e\n\u003cli\u003eWang, Y. \u003cem\u003eet al.\u003c/em\u003e Superhydrophobic wearable sensor: fabrication, application, and perspective. \u003cem\u003eDiscov. Nano\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 176 (2024).\u003c/li\u003e\n\u003cli\u003eTaudt, C., Leyens, C., Hartmann, P. \u0026amp; Molotnikov, A. Darkfield-scattering surface analysis in powder-based AM-processes: analysis of a multiwavelength approach. in \u003cem\u003eLaser 3D Manufacturing XII\u003c/em\u003e (eds. Helvajian, H., Gu, B. \u0026amp; Chen, H.) 16 (SPIE, San Francisco, United States, 2025). doi:10.1117/12.3043198.\u003c/li\u003e\n\u003cli\u003eTang, W. \u003cem\u003eet al.\u003c/em\u003e From finger friction to brain activation: Tactile perception of the roughness of gratings. \u003cem\u003eJ. Adv. Res.\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 129\u0026ndash;139 (2020).\u003c/li\u003e\n\u003cli\u003ePark, J., Yoo, H.-S., Hong, K. H. \u0026amp; Kim, E. Knitted fabric properties influencing coolness to the touch and the relationship between subjective and objective coolness measurements. \u003cem\u003eText. Res. J.\u003c/em\u003e \u003cstrong\u003e88\u003c/strong\u003e, 1931\u0026ndash;1942 (2018).\u003c/li\u003e\n\u003cli\u003eDegraen, D., Piovarči, M., Bickel, B. \u0026amp; Kr\u0026uuml;ger, A. Capturing Tactile Properties of Real Surfaces for Haptic Reproduction. in \u003cem\u003eThe 34th Annual ACM Symposium on User Interface Software and Technology\u003c/em\u003e 954\u0026ndash;971 (ACM, Virtual Event USA, 2021). doi:10.1145/3472749.3474798.\u003c/li\u003e\n\u003cli\u003eSousa, E. \u003cem\u003eet al.\u003c/em\u003e Tactile and visual perception of plastic textures for car interiors: Psychophysical and affective dimensions. \u003cem\u003eInt. J. Ind. Ergon.\u003c/em\u003e \u003cstrong\u003e92\u003c/strong\u003e, 103369 (2022).\u003c/li\u003e\n\u003cli\u003eSutu, A., Meftah, E.-M. \u0026amp; Chapman, C. E. Physical determinants of the shape of the psychophysical curve relating tactile roughness to raised-dot spacing: implications for neuronal coding of roughness. \u003cem\u003eJ. Neurophysiol.\u003c/em\u003e \u003cstrong\u003e109\u003c/strong\u003e, 1403\u0026ndash;1415 (2013).\u003c/li\u003e\n\u003cli\u003eChiba, T., Kuroda, S. \u0026amp; Yamaguchi, M. Modeling the relationship between tactile sensation and physical properties of synthetic leather. \u003cem\u003eJ. Ind. Text.\u003c/em\u003e \u003cstrong\u003e50\u003c/strong\u003e, 346\u0026ndash;363 (2020).\u003c/li\u003e\n\u003cli\u003eWu, J., Li, N., Liu, W., Song, G. \u0026amp; Zhang, J. Experimental Study on the Perception Characteristics of Haptic Texture by Multidimensional Scaling. \u003cem\u003eIEEE Trans. Haptics\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 410\u0026ndash;420 (2015).\u003c/li\u003e\n\u003cli\u003eLee, H.-T., Cheon, H.-R., Lee, S.-H., Shim, M. \u0026amp; Hwang, H.-J. Publisher Correction: Risk of data leakage in estimating the diagnostic performance of a deep-learning-based computer-aided system for psychiatric disorders. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 2172 (2024).\u003c/li\u003e\n\u003cli\u003eJames, G., Witten, D., Hastie, T., Tibshirani, R. \u0026amp; Taylor, J. Linear Model Selection and Regularization. in \u003cem\u003eAn Introduction to Statistical Learning\u003c/em\u003e 229\u0026ndash;288 (Springer International Publishing, Cham, 2023). doi:10.1007/978-3-031-38747-0_6.\u003c/li\u003e\n\u003cli\u003eSkedung, L., Harris, K. L., Collier, E. S. \u0026amp; Rutland, M. W. The finishing touches: the role of friction and roughness in haptic perception of surface coatings. \u003cem\u003eExp. Brain Res.\u003c/em\u003e \u003cstrong\u003e238\u003c/strong\u003e, 1511\u0026ndash;1524 (2020).\u003c/li\u003e\n\u003cli\u003eIn Hwan Sul, Kyung Hwa Hong, Shim, H., \u0026amp; Tae Jin Kang. Surface Roughness Measurement of Nonwovens Using Three-dimensional Profile Data. \u003cem\u003eText. Res. J.\u003c/em\u003e \u003cstrong\u003e76\u003c/strong\u003e, 828\u0026ndash;834 (2006).\u003c/li\u003e\n\u003cli\u003eFehlberg, M., Monfort, E., Saikumar, S., Drewing, K. \u0026amp; Bennewitz, R. Perceptual Constancy in the Speed Dependence of Friction During Active Tactile Exploration. \u003cem\u003eIEEE Trans. Haptics\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 957\u0026ndash;963 (2024).\u003c/li\u003e\n\u003cli\u003eYeo, J. C. \u003cem\u003eet al.\u003c/em\u003e Wearable Mechanotransduced Tactile Sensor for Haptic Perception. \u003cem\u003eAdv. Mater. Technol.\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 1700006 (2017).\u003c/li\u003e\n\u003cli\u003eTadesse, M. G., Chen, Y., Wang, L., Nierstrasz, V. \u0026amp; Loghin, C. Tactile Comfort Prediction of Functional Fabrics from Instrumental Data Using Intelligence Systems. \u003cem\u003eFibers Polym.\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 199\u0026ndash;209 (2019).\u003c/li\u003e\n\u003cli\u003eSkedung, L. \u003cem\u003eet al.\u003c/em\u003e Feeling Small: Exploring the Tactile Perception Limits. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 2617 (2013).\u003c/li\u003e\n\u003cli\u003eNuriyev, M. N. \u0026amp; Jafarova, A. M. Development of a method for the combined control of the hardness of winding textile package. \u003cem\u003eEUREKA Phys. Eng.\u003c/em\u003e 74\u0026ndash;84 (2022) doi:10.21303/2461-4262.2022.002237.\u003c/li\u003e\n\u003cli\u003eTurl, L. H. The Measurement of Tearing Strength of Textile Fabrics. \u003cem\u003eText. Res. J.\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 169\u0026ndash;176 (1956).\u003c/li\u003e\n\u003cli\u003eHu, Y., Hu, J., Zhao, Q., Ding, X. \u0026amp; Yang, X. Relationship between tactual roughness judgment and surface morphology of fabric by fingertip touching method. \u003cem\u003eFibers Polym.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 1024\u0026ndash;1031 (2013).\u003c/li\u003e\n\u003cli\u003eKrstajic, D., Buturovic, L. J., Leahy, D. E. \u0026amp; Thomas, S. Cross-validation pitfalls when selecting and assessing regression and classification models. \u003cem\u003eJ. Cheminformatics\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 10 (2014).\u003c/li\u003e\n\u003cli\u003eShrestha, N. Detecting Multicollinearity in Regression Analysis. \u003cem\u003eAm. J. Appl. Math. Stat.\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 39\u0026ndash;42 (2020).\u003c/li\u003e\n\u003cli\u003eSimunovic, G. \u003cem\u003eet al.\u003c/em\u003e Surface roughness assessing based on digital image features. \u003cem\u003eAdv. Prod. Eng. Manag.\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 93\u0026ndash;104 (2016).\u003c/li\u003e\n\u003cli\u003eKarim, A. K. M. R., Prativa, S. \u0026amp; Likova, L. Perception and appreciation of tactile objects: The role of visual experience and texture parameters (JPI-first). \u003cem\u003eJ. Percept. Imaging\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"tactile perception, smoothness/roughness perception, mechanical properties, regression model","lastPublishedDoi":"10.21203/rs.3.rs-7046348/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7046348/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAccurately predicting human perception of tactile roughness remains challenging because previous models often used limited mechanical properties, small sample sizes, and insufficient validation methods. To address these limitations, we developed a predictive model integrating multidimensional mechanical properties and subjective evaluations of tactile perception, using 50 commercially available synthetic fiber samples, including polyester, spandex, nylon, and their blends. Twelve mechanical properties were measured across four categories: geometric roughness, frictional force, hardness, and tensile strength. Tactile perception of smoothness/roughness was evaluated by 37 participants using a 5-point scale, with lower values indicating smoother textures and higher values indicating rougher textures. Correlation analysis identified kinetic friction coefficient (KF, ρ = -0.67), arithmetic mean roughness (Ra, ρ\u0026thinsp;=\u0026thinsp;0.44), mean width of profile elements (RSm, ρ\u0026thinsp;=\u0026thinsp;0.42), maximum load (ML, ρ = -0.41), and root mean square slope (Rdq, ρ\u0026thinsp;=\u0026thinsp;0.31) as key predictors. Among six regression models, Gaussian process regression showed the highest predictive accuracy (cross-validated R\u0026sup2; = 0.71). Comparisons between non-cross-validated and cross-validated results revealed substantial performance drops in cross-validation, underscoring the risk of performance overestimation without rigorous validation. The proposed framework provides a robust, generalizable approach applicable to broader tactile dimensions, benefiting material evaluation, product development, and haptic technologies.\u003c/p\u003e","manuscriptTitle":"Predicting Human Tactile Smoothness/Roughness Perception from Multidimensional Mechanical Properties of Synthetic Fibers Using Machine Learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-12 14:58:36","doi":"10.21203/rs.3.rs-7046348/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-09-18T16:03:40+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-18T00:31:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"104638414109371665908500722834100176944","date":"2025-09-14T22:38:24+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-14T20:18:44+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"42583126045681022296799478876734802398","date":"2025-09-09T07:23:35+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-08T13:24:28+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"198158018837763258719759080917309671007","date":"2025-09-07T23:41:43+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"26863596813078658743784119603340914897","date":"2025-09-07T00:15:29+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"20991355784265273641906491789987759575","date":"2025-09-06T19:25:30+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-06T18:47:14+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-12T02:34:49+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-07T10:51:32+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-07-07T10:48:01+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"1bf1cebd-1818-4490-b8d8-b62ee0d6b94c","owner":[],"postedDate":"September 12th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":54414745,"name":"Physical sciences/Engineering"},{"id":54414746,"name":"Physical sciences/Materials science"}],"tags":[],"updatedAt":"2025-12-01T16:06:16+00:00","versionOfRecord":{"articleIdentity":"rs-7046348","link":"https://doi.org/10.1038/s41598-025-26294-5","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-11-27 15:57:11","publishedOnDateReadable":"November 27th, 2025"},"versionCreatedAt":"2025-09-12 14:58:36","video":"","vorDoi":"10.1038/s41598-025-26294-5","vorDoiUrl":"https://doi.org/10.1038/s41598-025-26294-5","workflowStages":[]},"version":"v1","identity":"rs-7046348","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7046348","identity":"rs-7046348","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}