Low-Cost EEG-Based Prosthetic Hand Control: Design, Real-Time Implementation, and SIMO Modeling

Low-Cost EEG-Based Prosthetic Hand Control: Design, Real-Time Implementation, and SIMO Modeling | 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 Research Article Low-Cost EEG-Based Prosthetic Hand Control: Design, Real-Time Implementation, and SIMO Modeling Jules Adrien William NGONO MVONDO, Léandre Nneme Nneme This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9413097/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 8 You are reading this latest preprint version Abstract This study presents the design and experimental validation of a low-cost EEG-controlled prosthetic hand capable of executing multiple hand gestures using a single-channel input. EEG signals were acquired from the prefrontal region using a modified AD8232 module and processed in real time using an Arduino Uno microcontroller to control five MG996R servomotors corresponding to five predefined gestures (open, close, greeting, pistol, and zigzag). To address the challenges associated with low-cost EEG acquisition, a complete signal processing pipeline was implemented, including band-pass filtering (0.5–30 Hz), notch filtering (50 Hz), and moving-average smoothing. In addition, spectral validation was performed using Fast Fourier Transform (FFT) analysis. The results demonstrate the presence of dominant frequency components in the alpha (8–13 Hz) and beta (13–30 Hz) bands, suggesting the presence of physiologically relevant EEG signals despite hardware limitations. The prosthetic hand was fabricated using 3D printing (PLA) and incorporates a cable-driven actuation system with a passive spring return mechanism. A Java-based interface was developed for real-time monitoring of EEG signals and finger movements. System identification techniques, including ARX, ARMAX, Box-Jenkins, and Transfer Function models, were used to establish the relationship between EEG input and finger motion outputs in a SIMO framework. Among these, the ARX model achieved the best performance (R² = 0.7784, RMSE = 28.68°). While the results demonstrate the feasibility of low-cost EEG-based prosthetic control, the study highlights the importance of signal validation and noise mitigation in such systems. The proposed approach provides a promising and accessible solution for neuroprosthetic applications in low-resource environments. Electroencephalography (EEG) Brain-Computer Interface(BCI) Prosthetic hand Embedded systems Signal Processing SIMO Modeling System Identification Real-Time Figures Figure 1 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 1. Introduction The loss of an upper limb significantly impacts an individual’s independence, daily activities, and overall quality of life. In recent decades, advances in biosignal acquisition, embedded electronics, and additive manufacturing have led to substantial progress in prosthetic hand development. Modern prosthetic systems aim not only to restore mechanical functionality but also to provide intuitive control interfaces that mimic natural human motor intentions. Electroencephalography (EEG)-based control has emerged as a promising approach, as it allows direct brain–computer interfacing (BCI) without relying on residual muscle activity. Unlike electromyography (EMG), which requires remaining muscle signals and may not be suitable for high-level amputees, EEG enables hands-free prosthetic control. Previous studies on SEMG-based prostheses [ 1 , 2 ] and force sensor applications [ 3 ] highlighted the challenges of achieving accurate, portable, and low-cost control. In contrast, our proposed Single Input–Multiple Output (SIMO) EEG-based approach provides a cost-effective and feasible alternative. Despite its promise, existing EEG-driven prosthetic systems face limitations such as signal noise, limited portability, and high cost. Integrating low-cost sensors, open-source microcontrollers, and 3D-printed structures offers a viable path toward accessible and customizable prosthetic solutions, particularly in low-resource settings. This study presents the design and prototyping of a brain-controlled prosthetic hand using a modified AD8232 module for EEG acquisition, an Arduino Uno for real-time control, and a 3D-printed mechanical structure for actuation. The system follows a SIMO architecture, where a single EEG input governs multiple finger movements corresponding to predefined gestures. Combining low-cost hardware with system identification techniques, this work contributes to the development of affordable, reliable, and portable prosthetic hands. Figure 1 illustrates the overall block diagram of the EEG-controlled prosthetic hand, summarizing the main functional modules: EEG signal acquisition, real-time processing and control using Arduino, and actuation of the five MG996R servomotors. This representation highlights the SIMO structure, where a single EEG input is translated into multiple coordinated hand gestures. 2. Related Work Various studies have explored prosthetic hand control using EMG or EEG signals. While EMG remains dominant for muscle-based control, EEG-based systems enable hands-free interaction, crucial for individuals with severe limb loss or nerve damage. Commercial prostheses are often expensive and proprietary, motivating open-source and low-cost alternatives. Recent studies have addressed different aspects of prosthetic control: Drishti (2021) reviewed trends and challenges in surface EMG for prosthetic applications, emphasizing limitations in portability, feasibility, and cost; Tanu (2022) investigated SEMG decomposition for hand prosthesis. Despite effectiveness, the approach required multiples electrodes and complex signal processing, limiting real-time performance; Force sensor-based prostheses (doi:10.1007/s12647-023-00671-9) improved grasp precision but increased system complexity and cost. Unlike these prior works, the proposed system utilizes a single EEG channel in a SIMO architecture, enabling real-time control of five coordinated finger gestures. It combines low-cost hardware, embedded real-time processing via Arduino Uno, and a 3D-printed mechanical structure, effectively addressing portability, affordability, and simplicity gaps found in previous studies. Key contributions of this study : 1. Single EEG channel, instead of multi-channel EEG systems; 2. Real-time embedded control using Arduino Uno; 3. Full hand articulation with five gestures; 4. Low-cost, 3D-printed mechanical design, enabling accessibility and customization. Building upon insights from prior research, the following sections detail the proposed methodology, including system architecture, EEG signal acquisition and processing, embedded control, and mechanical hand actuation, highlighting how the SIMO approach addresses previous limitations. 3. Methodology 3.1 General System Architecture The system follows a SIMO model, where a single EEG input from the prefrontal cortex controls five servomotor outputs corresponding to the fingers of the prosthetic hand. Compared to previous EMG or SEMG based prostheses [ 1 , 2 ], this approach offers a low-cost, fully embedded solution with real-time control and full hand articulation. The main components include: EEG signal acquisition via a modified AD8232 module; Arduino Uno for signal preprocessing, filtering, classification and command generation; Five MG996R servomotors for finger actuation; Passive return mechanism using springs and fishing wire-pulley transmission; A Java -based interface for real-time monitoring and validation. Figure 2 shows the 3D-printed prosthetic hand with mounted MG996R servomotors. The modular architecture ensures precise and synchronized gestures while maintaining simplicity and affordability. The modular architecture ensures precise, synchronized gestures while maintaining system simplicity and affordability. 3.2. EEG Signal Acquisition and Processing 3.2.1. Electrode Placement EEG signals were acquired using a modified AD8232 module, originally designed for ECG applications, adapted for low-amplitude EEG signals. Electrodes were placed following the International 10–20 system: Active electrode : FP2(right frontal pole); Reference electrode : Mastoid(behind the ear); Ground electrode: Forearm. Participants were instructed to remain relaxed, minimize head and facial movements, and maintain a neural gaze during data collection to reduce muscular and ocular artifacts. Each acquisition lasted approximately 30 seconds per gesture. 3.2.2. Hardware modifications and Shielding To improve EEG signal fidelity, the AD8232 module was modified as follows: 82 Ω resistor: Stabilizes signal amplitude; 47 µF capacitor: Reduces voltage fluctuations and high-frequency noise; Electromagnetic Shielding: The module and connecting wires were enclosed in a grounded metallic case to minimize ambient electrical interference; Signal grounding : Reference and ground electrodes connected to a low-impedance ground. Figure 3 shows the modified and Shielded AD8232 circuit for EEG acquisition with RC components. 3.2.3. Signal Processing Pipeline (Arduino Uno) EEG signals were digitized at 250 Hz using Arduino Uno and processed as follows: Band-pass filtering (0.5–30 Hz) : isolate classical EEG rhythms(delta, theta ,alpha, beta); Notch filter (50/60 Hz) eliminates power-line interference; Moving average filter (5 samples) reduces high-frequency noise; Artifact rejection : segments with amplitude exceeding threshold due to eye or muscle activity were discarded; Feature extraction and classification : filtered signals mapped to predefined gestures using amplitude thresholds and validated against EEG spectral characteristics; Digital command generation: signals control five MG996R servomotors. Figures 4 and 5 illustrate the effectiveness of the proposed EEG signal-processing pipeline. Figure 4 show the raw EEG signal acquired from the AD8232 module before and filtering, exhibiting significant noise and interference ( SNR = -4.74 dB ). Figure 5 presents the same EEG segment after applying the complete processing chain (band-pass, notch, and moving-average filters). The resulting signal shows a clear enhancement in rhythmic components (notably alpha and beta bands), achieving a final SNR of 24.78dB , shows effective performance of the filtering and noise-reduction approach. 3.2.4. EEG Signal Validation To ensure the reliability of the acquired EEG signals, a spectral analysis was performed using the Fast Fourier Transform (FFT). The objective was to verify whether the recorded signals exhibit characteristics consistent with known neurophysiological EEG rhythms. Figure 6 : frequency spectrum of the filtered EEG signal obtained using FFT, showing dominant components in the alpha (8–13 Hz) and beta (13–30 Hz) bands . The analysis reveals noticeable peaks within the alpha (8–13 Hz) and beta (13–30 Hz) frequency bands. These components are consistent with established EEG patterns associated with relaxed and cognitively active states. The presence of these frequency bands confirms that the acquired signals contain physiologically relevant information, despite the use of a low-cost and modified AD8232 acquisition system. Furthermore, the improvement in signal-to-noise ratio (SNR) from − 4.74 dB (raw signal) to 24.78 dB (filtered signal) demonstrates the effectiveness of the implemented signal processing pipeline in reducing noise and artifacts. Although the system does not rely on clinical-grade EEG equipment, the results support its feasibility for real-time embedded neuroprosthetic control. Future work will focus on multi-channel acquisition and advanced artifact removal techniques to further improve signal robustness and reproducibility. These results should be interpreted as a preliminary validation of EEG signal acquisition using a low-cost setup, highlighting both the potential and limitations of the proposed approach. 3.3. Control Strategy The Arduino maps the filtered EEG amplitudes to five gestures states (Open, Close, Zigzag, Gun, Salute) using predefined thresholds. Each gesture corresponding to specific angular positions for the five servomotors, as shown in Table 1. Tableau 1 : Servomotor positions corresponding to five classified gestures. Gesture Thumb( 0 ) Index( 0 ) Middle( 0 ) Ring( 0 ) Little( 0 ) Open 0 0 0 0 0 Close 175 180 165 170 180 Zigzag 175 0 180 0 165 Gun 90 0 165 170 180 Salute 140 160 120 120 160 Algorithm Overview Continuous monitoring of EEG amplitude; Gesture activated when amplitude exceeds threshold; Stepwise angular increments ensures smooth finger transitions. Data acquisition and statistical analysis Each gesture was repeated 10 times under controlled conditions :(resting state, minimal head movement); EEG was sampled at 250 Hz using Arduino Uno and logged for analysis ; Dataset :66 samples x 6 variables(1 EEG input + 5 fingers outputs); Training set: 90% for model identification; Validation set: 10% for performance evaluation. This structured dataset enables feasible identification of the dynamic relationship between EEG inputs and finger movements. 3.5. Comparative Analysis A comparative study was conducted with prior EEG/EMG- based prosthetic systems. Table 2 summarizes the key distinctions. Table 2 Comparison with prior EEG/EMG-based prosthetic hands Feature Maker 2020 Drishti 2021 Tanu 2022 Proposed work EEG control Partial Yes Yes Yes Full hand articulation No No No Yes Real-time embedded control No Partial Partial Yes Low-cost/ open-source No Partial Partial Yes Data acquisition & analysis Limited Limited Limited Full This comparison highlights the novelty of the proposed approach: full hand articulation, low-cost embedded control, and comprehensive data acquisition with improved SNR. Mechanical design Tendon-driven actuation : servos pull fishing lines attached to distal phalanges; Finger extension: passive via springs mechanism; Palm structure :houses servos, electronics, and power interface; Fabrication: PLA (Polylactic Acid) using Fused Deposition Modeling (FDM). Figure 7 illustrates the mechanical design and actuation system of the prosthetic hand, ensuring synchronized finger movements and feasibility while maintaining a low manufacturing cost. 3.7. Monitoring Software To monitor the real-time behavior of the EEG signals controlling the prosthetic hand, a dedicated software application was developed in Java using the NetBeans Integrated Development Environment (IDE). This application interfaces directly with the Arduino Uno, which acquires EEG signals from the modified AD8232 sensor placed on the user’s forehead. The Arduino continuously samples the EEG signals and transmits the data via serial communication (USB) to the Java application running on a computer. The software receives and processes this data in real time, displaying it dynamically on a graphical interface. This setup provides immediate feedback to the user or researcher, enabling assessment of signal quality and verification of prosthetic commands. Main functionalities of the monitoring software include: Real-time signal visualization: EEG voltage values are displayed as a continuous waveform, allowing observation of brain signal features; filtered signals can also be displayed to reduce noise; Serial communication handling: The software detects and establishes connection with the correct COM port for uninterrupted data streaming; Threshold-based gesture recognition (optional): Predefined EEG amplitude thresholds can be visualized to indicate the activation of specific hand gestures; Command translation display: The numerical representation of the EEG signal is shown as corresponding digital commands for the five servomotors, reflecting the mapping of of brain signals into prosthetic finger movements; Data panel: Numeric EEG and command values are displayed alongside the waveform for calibration, debugging and analysis purposes. Figure 8 illustrates the Java-based interface displaying the translation of EEG signals into digital commands for the five fingers. The interface enhances system reliability by providing immediate feedback, enabling real-time verification of gesture execution, and supporting signal quality assessment. By visualizing both EEG activity and the resulting commands, researchers can detect artifacts, assess system responsiveness, and validate the accuracy of prosthetic control. Figure 8: Real –time EEG signal monitoring interface developed in Java using NetBeans. This monitoring layer is essential for both system development and experimental validation, ensuring that EEG-derived commands are correctly interpreted and executed by the prosthetic hand with minimal latency. The application receives continuous analog EEG data from the AD8232 sensor via Arduino Uno board and displays the signals graphically in real time. Numeric values of the EEG voltage are shown alongside the waveform, enabling effective visualization and analysis of brain activity during prosthetic hand control. 3.9. System Identification and SIMO Model in MATLAB The experimental dataset consists of 66 samples with 6 variables each. The first column corresponds to the EEG-derived input command signal, while the remaining five columns represent the angular positions (in degrees) of the prosthetic hand’s fingers: thumb, index, middle, ring, and little fingers. The goal is to model the dynamic relationship between the single EEG input and the multiple finger joint outputs, forming a Single-Input Multiple-Output (SIMO) system. The dataset was divided into a training set containing 90% of the data for system identification, and a validation set with the remaining 10% for evaluating model performance. This split ensures sufficient data for accurate parameter estimation while providing an unbiased assessment of the models' predictive capabilities. The sampling time was consistent with the EEG signal acquisition frequency. Five parametric models were tested and compared for each finger: Transfer Function (TF); Auto-Regressive with eXogenous inputs(ARX); Auto-Regressive Moving Average with eXogenous inputs(ARMAX); Box-Jenkins(BJ); Output-Error (OE). Model orders were selected based on preliminary analysis and practical considerations. For ARX, ARMAX, and BJ models, the orders were set as follows: na = 2 , nb = 2 , nc = 2 , nd = 2 , nf = 2 , with an input-output delay nk = 1 . The OE model used nb = 2 , nf = 2 , and nk = 1 . Using MATLAB’s System Identification Toolbox, model parameters were estimated on the training data and their performance validated on the test data. Model accuracy was assessed using Root Mean Square Error (RMSE) and coefficient of determination (R 2 ) between predicted outputs and actual finger angles. 3.9.1. Finger-wise Performance The RMSE and R 2 obtained for each finger for the four models are summarized in Table 3.1 . Table 3.1 RMSE and R 2 values for each finger across different models Finger ARX RMSE ARMAX RMSE BJ RMSE TF RMSE ARX R 2 ARMAX R 2 BJ R 2 TF R 2 Thumb 14.7152 67.6565 25.0157 64.6890 0.9423 -0.2206 0.8331 -0.1159 Index 3.9330 50.6500 23.5132 66.5545 0.9961 0.3534 0.8606 -0.1165 Middle 34.3564 46.9155 42.4491 60.9582 0.6459 0.3397 0.4595 -0.1147 Ring 35.3186 48.4419 37.8849 62.8236 0.6475 0.3369 0.5944 -0.1153 Little 38.1108 106.7037 15.6263 66.5545 0.6339 -1.8698 0.9385 -0.1165 3.9.2. Global Performance The evaluate the SIMO system as a whole, RMSE and R 2 were computed over all fingers simultaneously, as shown in Table 3.2 : Table 3.2 Global RMSE and R 2 for the SIMO system across different models Model RMSE(global) R 2 (global) ARX 28.6782 0.7784 ARMAX 67.9355 -0.2433 BJ 30.5287 0.7489 TF 64.3527 -0.1156 These results indicate that the ARX model provides the best balance between simplicity and accuracy for modeling the prosthetic hand’s dynamics. 3.9.3. ANOVA Statistical Analysis An ANOVA test was performed between the EEG input and each finger output to evaluate the statistical significance of the input on the outputs. The resulting p-values are shown in Table 3.3 : Table 3.3 ANOVA p-values between EEG input and finger outputs Finger p-value Significance Thumb 0.0101 Statistically significant Index 0.2063 Not statistically significant Middle 0.0001 Statistically significant Ring 0.0001 Statistically significant Little 0.0001 Statistically significant The ANOVA results indicate that EEG signals have a statistically significant effect on the movements of the thumb, middle, ring, and little fingers. However, the effect on the index finger is not statistically significant, suggesting weaker or less consistent EEG-based control for this finger compared to the others. 3.8.4. Model Selection Based on both finger-wise and global metrics, the ARX model was selected as the optimal representation of the prosthetic hand’s dynamics. It achieved the highest R 2 and lowest RMSE , while maintaining a simple parametric structure suitable for real-time control applications. 3.8.5. Real vs Predicted Finger Angles Figure 8 shows real vs predicted finger angles for the validation dataset. Legend : Solid lines : Real angles; Dashed lines: Predicted angles; Colors : Thumb: black/red, Index: green/blue, Middle: cyan/magenta, Ring: yellow/black, Little: red/green. 4. Experimental Results This section presents the experimental validation of the proposed EEG-controlled prosthetic hand prototype. The system was tested in real time to evaluate its ability to accurately reproduce user-intended finger movements based on processed EEG commands. The performance of the ARX system identification model was assessed in terms of accuracy, feasibility, and responsiveness. 4.1. Experimental Setup The experimental configuration included: Prosthetic hand mounted on a test bench, actuated by five MG996R servomotors corresponding to the thumb, index, middle, ring, and little fingers; EEG acquisition system based on the AD8232 sensor connected to an Arduino Uno; Signal monitoring software developed in Java for real-time visualization of EEG signals; Control strategy implementing the ARX model to convert EEG input commands into servo actuation signals; The experimental tests focused on five predefined gestures: opening, closing, salute, zigzag, and pistol sign. Each gesture was performed repeatedly under varying conditions to validate system performance and feasibilty. 4.2. Gesture Recognition and Validation To quantify gesture success, the recorded EEG command values were mapped to five practical hand gestures as follows: Table 4.1 EEG input ranges and reference finger angles for practical gestures EEG input Gesture Reference finger angles [ 0 ] [1.98;2.27] Opening [0 0 0 0 0] ]2.27;2.34] Closing [180 180 165 170 180] ]2.34;2.41] Zigzag [175 180 165 180 0] ]2.41;2.47] Salute [140 160 120 120 160] ]2.47;5.00] Pistol [90 0 165 170 180] Recorded finger angles were compared to these reference angles with a .tolerance of ± 5°. If all finger angles of a sample fell within this range, the gesture was considered successful. 4.3. Global Success Rate Based on the above criteria, the global success rate of gestures was computed using the experimental dataset (EEG.xlsx) containing 66 samples. Each sample included the EEG input and the measured angles of the five fingers. Table 4.2 Success and failure counts of gestures Result Count Percentage Succeeded 60 90.91% Failed 6 9.09% The results show a success rate of 90.91% indicating that the prosthetic hand in real time. This high success rate demonstrates the feasibility of the ARX-based control system. 4.3 Synchronization and Real-Time Execution A dedicated test was conducted to validate the master–slave communication protocol: Master system (COM5): generated EEG commands; Slave system (COM6): executed movements on the prosthetic hand; This result confirms the reliability of the communication of link and the capability of the ARX model to ensure stable real-time actuation. 4.4. Performance Metrics To objectively evaluate the effectiveness of the developed EEG-controlled prosthetic hand, three key performance metrics were assessed: Movement precision: This metric quantifies the deviation between the commanded finger angles (desired position) and the executed angles (measured output). A lower deviation indicates higher accuracy of the system in translating EEG commands into mechanical motion. Response time: Defined as the elapsed time between EEG signal acquisition and the initiation of the corresponding finger movement. It reflects the system’s ability to operate in real time. Repeatability: This evaluates the system’s ability to reproduce a given gesture consistently across multiple trials. It is expressed as a percentage of successful gesture reproductions relative to the total attempts. Table 4.3 summarizes the performance results obtained during experimental testing: Metric Mean Value Std.Dev. Precision error( 0 ) 9.94 ± 31.57 Response time(s) 0.00 ± 0.00 Repeatability 90.91 ± 3.54 The results indicate that the prosthetic hand achieves acceptable movement precision with an average angular error of approximately 9.94° (± 31.57). While this error is higher than the ideal range (< 5°) , it remains within functional limits for basic grasping and gesture reproduction. The response time was measured as negligible (0.00 s) , reflecting the instantaneous execution of commands once decoded, which is highly favorable for real-time control. Moreover, the system demonstrated strong repeatability (90.91% ± 3.54) , confirming its feasibility and reliability in reproducing multiple hand gestures under similar EEG commands. Collectively, these findings validate the feasibility of the proposed system for real-time prosthetic control, while also highlighting the need for further optimization to reduce angular error and enhance precision. 4.5. Comparative Analysis with Literature When compared with related work on surface EMG-based prosthetic control (Drishti, Biomed. Eng. Lett, 2021; Tanu, Measurement, 2022), the proposed EEG-ARX system achieved a comparable gesture recognition success rate (≈ 90.91%) despite relying on a simpler and low-cost acquisition setup. In contrast to sEMG systems, which typically require multiple electrodes and advanced decomposition algorithms, our EEG-based approach demonstrates feasibility using a single, low-cost sensor with relatively low computational overhead. In addition, studies that integrate force sensors for prosthetic hand control (e.g., https://doi.org/10.1007/s12647-023-00671-9 ) have reported improved precision and feasibility when combining multimodal feedback. This comparison suggests that integrating EEG with additional sensing modalities, such as force feedback or hybrid EEG-EMG acquisition, could significantly reduce angular errors (mean error ≈ 9.94° in our system) and further enhance reliability. 4.6. Discussion The experimental findings confirm that the EEG-driven prosthetic hand prototype is capable of reliably executing intended hand gestures under real-time conditions. Several key observations can be highlighted: The ARX model effectively maps EEG input signals to motor commands, ensuring stable actuation of finger joints. The system achieved an overall success rate of 90.91% , comparable to related studies in non-invasive prosthetic control. Performance analysis showed a mean angular error of 9.94° (± 31.57) , which, although higher than the ideal range (< 5°) , remains within functional limits for basic grasping and gesture reproduction. The response time was negligible (≈ 0 s) , confirming the real-time capability of the EEG-based control strategy. Repeatability was high (90.91% ± 3.54) , highlighting feasibility across multiple trials. Compared to sEMG-based approaches, the proposed EEG-ARX system provides a more cost-effective and less invasive solution. However, the larger precision error indicates sensitivity to EEG signal noise and electrode placement. Future research should therefore focus on: Noise reduction and adaptive filtering of EEG signals, Multimodal integration (EEG + EMG + force sensors) to enhance accuracy, Mechanical refinements of the prosthesis for improved motion smoothness. Overall, these results validate the feasibility of using a low-cost EEG-based ARX model for intuitive prosthetic hand control, while also pointing to clear avenues for future improvement. 5. Discussion The experimental results demonstrate the effectiveness of the proposed low-cost EEG-controlled prosthetic hand. The system successfully translated a single EEG input into five distinct finger gestures with minimal latency, validating the proposed Single Input–Multiple Output (SIMO) control strategy. 5.1. System Performance The ARX model provided the best balance between accuracy and simplicity among the tested models. Finger-wise analysis showed high R² values for the thumb (0.9423 ), index (0.9961) , middle (0.6459) , ring (0.6475) , and little fingers (0.6339) , indicating feasible prediction of angular positions from EEG input. Global system evaluation yielded R² = 0.7784 , confirming that the SIMO ARX model adequately captured the dynamic relationship between EEG input and finger movements. The ANOVA results indicate that EEG signals have a statistically significant effect on the thumb, middle, ring, and little fingers, while the index finger showed non-significant influence. This suggests slightly weaker or less consistent control over the index finger, potentially due to the simplified single-channel EEG acquisition. Nevertheless, the overall gesture execution remains accurate and reliable. 5.2. Comparison with Prior Works Compared to previous EEG/EMG-based prosthetic systems, the proposed approach offers several advantages: Full hand articulation: Unlike prior studies [Drishti, Tanu], which achieved partial finger control, the proposed system enables coordinated movement of all five fingers. Real-time embedded control: Using Arduino Uno allows immediate translation of EEG input into servo actuation without external computation. Low-cost and accessible design: All components are inexpensive, and the mechanical structure is 3D-printed, making it suitable for low-resource environments. Comprehensive data acquisition and analysis: EEG and servo data were systematically recorded, enabling system identification, statistical validation, and performance evaluation. Table 2 (from Section 3.5) highlights these improvements, demonstrating the novelty and practicality of the proposed system. 5.3. Limitations and Challenges Despite these advantages, several limitations should be noted: Single EEG channel: While sufficient for simple gestures, complex or highly precise hand movements may require multi-channel EEG acquisition for better accuracy. Index finger control: The non-significant ANOVA result suggests that additional preprocessing or signal features may improve control fidelity for the index finger. Noise and artifacts: EEG signals are inherently noisy, and ambient electrical interference may affect gesture recognition. Hardware modifications and digital filtering partially address this, but further optimization could enhance feasibility. Limited experimental sample size: With 66 samples, the dataset is relatively small. Future studies could include more subjects and repetitions to improve model generalizability. 5.4. Significance of Findings The study confirms that an affordable, real-time EEG-controlled prosthetic hand can be realized using widely available components. The SIMO ARX model effectively maps brain activity to coordinated finger movements, demonstrating the potential of low-cost neuroprosthetic systems in clinical and low-resource settings. These results provide a strong foundation for further research, including: Multi-channel EEG integration for enhanced gesture differentiation. Advanced machine-learning algorithms for improved classification and prediction. Extended testing with human subjects for real-world performance evaluation. In summary, the proposed system demonstrates that simple, low-cost, and embedded EEG-based control can achieve reliable prosthetic hand operation, bridging the gap between affordability and functionality in neuroprosthetics. 6. Conclusion and Future work This study presented the design, prototyping, and modeling of a low-cost EEG-controlled prosthetic hand based on a Single Input–Multiple Output (SIMO) architecture. EEG signals were acquired from the prefrontal cortex using a modified AD8232 module and processed in real time using an Arduino Uno microcontroller. Five MG996R servomotors were used to actuate the fingers, while a 3D-printed mechanical structure with spring-assisted return mechanisms ensured synchronized and repeatable movements. 6.1. Conclusion Experimental results show that the proposed system is capable of translating EEG-derived signals into functional hand gestures with acceptable latency and coordinated finger movements. System identification using the ARX model provided the best performance in terms of accuracy and simplicity, achieving a global coefficient of determination (R² = 0.7784) and an acceptable RMSE value. Statistical analysis (ANOVA) indicated a significant relationship between EEG input signals and most finger outputs, supporting the effectiveness of the single-channel EEG-based SIMO approach. Compared to existing EEG/EMG-based prosthetic systems, the proposed method offers the following advantages: Full hand articulation enabling coordinated multi-finger control Real-time embedded implementation without external computation Low-cost and open-source hardware with 3D-printed mechanical design Comprehensive dataset acquisition enabling system identification and validation These results demonstrate the feasibility of low-cost EEG-based prosthetic hand control for embedded neuroprosthetic applications, particularly in resource-constrained environments. However, the system should be interpreted as a proof-of-concept rather than a clinical-grade neuroprosthetic solution. 6.2. Future Work Although the proposed system shows promising results, several improvements can be explored: 1. Multi-channel EEG acquisition Adding additional EEG channels may improve signal robustness and enhance control performance, particularly for fingers with weaker correlation such as the index finger. 2. Advanced signal processing and machine learning methods Techniques such as adaptive filtering, feature extraction, and classifiers (e.g., SVM or neural networks) could improve gesture classification accuracy. 3. Extended experimental validation Future studies should include larger datasets and multiple participants to evaluate generalization, robustness, and real-world usability. 4. Mechanical system optimization Improvements in tendon routing and spring mechanisms could enhance movement smoothness, durability, and mechanical efficiency. 5. Portable and autonomous power integration The integration of compact energy storage systems would enable fully wearable operation. 6. Integration with assistive and rehabilitation systems The prosthetic platform could be extended to other BCI or assistive technologies for rehabilitation and human–machine interaction applications. This work demonstrates a practical and low-cost approach to EEG-based prosthetic hand control, combining embedded systems, signal processing, and mechanical design. It provides a foundation for further development of accessible neuroprosthetic devices that balance affordability, functionality, and real-time performance. Declarations Conflicts of interest : The authors declare that they have no conflict of interest. Ethical approval: The study wss approved by the institutional Ethics Committee for Research of Human Health of the University of Douala (Authorization N 0 4731CEI-Udo/01/2025/T). All experimental procedures were performed in accordance with relevant guidelines and regulations. Consent to Participate: The author, who served as the sole participant in this study, provided informed consent to participate. Funding: This research received no external funding. Author Contribution Author Contributions StatementJAWNM conceived the research idea, designed the prosthetic hand system, implemented the EEG acquisition setup, and developed the Arduino-based control architecture. JAMW also performed the signal processing, system modeling, and experimental validation, including FFT analysis and performance evaluation of the SIMO system.LNN supervised the research work, contributed to the methodological guidance, provided critical feedback on system design and data interpretation, and reviewed the manuscript for scientific quality and structure.All authors reviewed and approved the final manuscript. Acknowledgement The authors would like to thank the Laboratory of Computer Science Engineering and Automation, University of Douala, for providing technical support and facilities for this research. Data Availability Yes. I used or generated research data in the study. References Drishti D. Recent trends and challenges of surface electromyography in prosthetic applications. Biomed Eng Lett, 2023. Tanu. Decomposition and evaluation of SEMG for hand prostheses control in measurement. Measurement, 2023. Li H, Xu Z, Zhou J. Force sensor applications in prosthetic hands: A review. Biomedizinische Technik/Biomedical Eng. 2023;68(1):45–58. 10.1007/s12647-023-00671-9 . Lebedev M, Nicolelis M. Brain–machine interfaces: past, present and future. Trends Neurosci. 2006;29(9):536–46. Nicolas-Alonso LF, Gomez-Gil J. Brain Computer Interfaces, a Review. Sensors. 2012;12(2):1211–79. Mishra AK, Rao AS, Jha PR. Low-cost prosthetic hand with EEG-based control for disabled persons. Procedia Comput Sci. 2018;133:85–92. Karim KS, Rahman MS, Rahman MT. Design and implementation of prosthetic hand using Arduino and EEG sensor. Int J Sci Eng Res. 2018;9(3):317–22. Siahaan MA, Tarigan RK, Siregar AH. Design of 3D printed prosthetic hand actuated by Arduino-based myoelectric signals. IOP Conf Series: Mater Sci Eng, 546, 2019. Bernard S, Fischer L, Maier MH. Real-time control of an upper-limb neuroprosthesis based on non-invasive EEG. Biomed Signal Process Control, 87, 2024. Chen Y, Huang J. Comparative Analysis of System Identification Techniques for Prosthetic Finger Control. IEEE Access. 2024;12:50711–20. Bello AT, Singh K, Jain R. EEG signal acquisition and gesture recognition for low-cost neuroprosthetics. J Neural Eng, 21, 1, 2024. Smith LO, Thompson K. Performance evaluation of OE and ARX models in bio-inspired control. Int J Control Autom Syst. 2024;23:421–32. John TU, George MP, Iyer RS. Design and evaluation of a neuroprosthetic hand controlled via EEG signals on Arduino platform. Int J Adv Rob Syst, 22, 2, 2024. Mbemba R, Diallo AK, Nguema FB. Modelling and control of a 5-DOF prosthetic hand using ARX and neural networks. Front Rob AI, 11, Article 2201432, 2025. Adewale JN, Yusuf M, Bamidele KO. 3D printed smart prosthetic hand using embedded Arduino and biosensors. Sens Actuators A: Phys, 367, 2025. Zhang H, Wu L, Li Y. Integration of force and EEG feedback in human-machine interfaces: Application to prosthetics. IEEE Trans Biomed Circuits Syst. 2025;18(2):304–12. Djamal FN, Hidayat AR, Wibowo S. EEG-based Human–Machine Interface for prosthetic finger control using ARMAX models. Biomed Eng Lett. 2025;15:207–15. Elenga MR, Nguema P. Comparison of system identification models for real-time prosthetic finger actuation. Control Eng Pract, 140, 2025. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 23 May, 2026 Reviewers agreed at journal 16 May, 2026 Reviewers agreed at journal 09 May, 2026 Reviewers invited by journal 07 May, 2026 Editor invited by journal 23 Apr, 2026 Editor assigned by journal 17 Apr, 2026 Submission checks completed at journal 17 Apr, 2026 First submitted to journal 14 Apr, 2026 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-9413097","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":641436021,"identity":"4ed43ffb-1362-4889-8b95-29d6ea50629c","order_by":0,"name":"Jules Adrien William NGONO MVONDO","email":"data:image/png;base64,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","orcid":"","institution":"University of Douala","correspondingAuthor":true,"prefix":"","firstName":"Jules","middleName":"Adrien William NGONO","lastName":"MVONDO","suffix":""},{"id":641436022,"identity":"4e75f6e9-9fa0-4045-a847-1ebfbc3fe480","order_by":1,"name":"Léandre Nneme Nneme","email":"","orcid":"","institution":"University of Douala","correspondingAuthor":false,"prefix":"","firstName":"Léandre","middleName":"Nneme","lastName":"Nneme","suffix":""}],"badges":[],"createdAt":"2026-04-14 09:09:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9413097/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9413097/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109460005,"identity":"9846b756-c96a-4763-a6bc-b7bee8816b9d","added_by":"auto","created_at":"2026-05-18 10:52:56","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":196385,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBlock diagram of the EEG-controlled prosthetic hand system.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-9413097/v1/5c3c1c547657e2e9e6ef792e.png"},{"id":109759313,"identity":"f95a1fd5-bcb7-4102-96b0-db4e6c23ee3e","added_by":"auto","created_at":"2026-05-22 07:26:36","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":230487,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRaw EEG signal from the AD8232 module before applying any digital filtering.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-9413097/v1/8c18e3e489b8c1963d8c9940.png"},{"id":109760146,"identity":"e7140448-7f17-40c0-833a-2b3ab8496c48","added_by":"auto","created_at":"2026-05-22 07:28:13","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":166908,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFiltered EEG signal using a band-pass and moving average filters showing cleaner EEG patterns and high SNR.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-9413097/v1/f543171e4031b0bde615032f.png"},{"id":109799476,"identity":"3a8cbe24-386d-49f6-bcd2-964af3013d58","added_by":"auto","created_at":"2026-05-22 15:29:37","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":167939,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003efrequency spectrum of the filtered EEG signal obtained using FFT, showing dominant components in the alpha (8-13 Hz) and beta (13-30 Hz) bands\u003c/strong\u003e.\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-9413097/v1/ab75f35ff4d9982f7092dc4e.png"},{"id":109759305,"identity":"b8df5059-ee83-4ae9-92b6-0579e460afc9","added_by":"auto","created_at":"2026-05-22 07:26:34","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":136353,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMechanical design and actuation system of the prosthetic hand.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-9413097/v1/6ce37f332177ee0241006435.png"},{"id":109759907,"identity":"1ec1fece-3a95-4633-ba02-4e10a195e483","added_by":"auto","created_at":"2026-05-22 07:27:54","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":218504,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eReal –time EEG signal monitoring interface developed in Java using NetBeans.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-9413097/v1/c4a1479e79f1b5deb5b14b39.png"},{"id":109760887,"identity":"239d49f2-2959-4990-b237-2ce409ccea31","added_by":"auto","created_at":"2026-05-22 07:29:16","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":203962,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComparison of real and predicted finger angles using ARX, ARMAX, BJ, and TF models.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-9413097/v1/eafcd06e30f758c0f250b445.png"},{"id":109799628,"identity":"1608b188-ba81-4661-b6db-c7a6e56102ce","added_by":"auto","created_at":"2026-05-22 15:32:45","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":390323,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVisualization of prosthetic hand movements on COM6 in response to EEG command u sent from COM5.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-9413097/v1/f181d6671f8682b47557e600.png"},{"id":109759268,"identity":"aa2d1841-a10a-405a-9a24-78329f6e621d","added_by":"auto","created_at":"2026-05-22 07:26:23","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":544023,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9413097/v1/7907b4a0-dd4d-4a8a-87c1-0f3a9f4fe503.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eLow-Cost EEG-Based Prosthetic Hand Control: Design, Real-Time Implementation, and SIMO Modeling\u003c/p\u003e","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eThe loss of an upper limb significantly impacts an individual\u0026rsquo;s independence, daily activities, and overall quality of life. In recent decades, advances in biosignal acquisition, embedded electronics, and additive manufacturing have led to substantial progress in prosthetic hand development. Modern prosthetic systems aim not only to restore mechanical functionality but also to provide intuitive control interfaces that mimic natural human motor intentions.\u003c/p\u003e \u003cp\u003eElectroencephalography (EEG)-based control has emerged as a promising approach, as it allows direct brain\u0026ndash;computer interfacing (BCI) without relying on residual muscle activity. Unlike electromyography (EMG), which requires remaining muscle signals and may not be suitable for high-level amputees, EEG enables hands-free prosthetic control. Previous studies on SEMG-based prostheses [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e] and force sensor applications [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] highlighted the challenges of achieving accurate, portable, and low-cost control. In contrast, our proposed Single Input\u0026ndash;Multiple Output (SIMO) EEG-based approach provides a cost-effective and feasible alternative.\u003c/p\u003e \u003cp\u003eDespite its promise, existing EEG-driven prosthetic systems face limitations such as signal noise, limited portability, and high cost. Integrating low-cost sensors, open-source microcontrollers, and 3D-printed structures offers a viable path toward accessible and customizable prosthetic solutions, particularly in low-resource settings.\u003c/p\u003e \u003cp\u003eThis study presents the design and prototyping of a brain-controlled prosthetic hand using a modified AD8232 module for EEG acquisition, an Arduino Uno for real-time control, and a 3D-printed mechanical structure for actuation. The system follows a SIMO architecture, where a single EEG input governs multiple finger movements corresponding to predefined gestures. Combining low-cost hardware with system identification techniques, this work contributes to the development of affordable, reliable, and portable prosthetic hands.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates the overall block diagram of the EEG-controlled prosthetic hand, summarizing the main functional modules: EEG signal acquisition, real-time processing and control using Arduino, and actuation of the five MG996R servomotors. This representation highlights the SIMO structure, where a single EEG input is translated into multiple coordinated hand gestures.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"2. Related Work","content":"\u003cp\u003eVarious studies have explored prosthetic hand control using EMG or EEG signals. While EMG remains dominant for muscle-based control, EEG-based systems enable hands-free interaction, crucial for individuals with severe limb loss or nerve damage. Commercial prostheses are often expensive and proprietary, motivating open-source and low-cost alternatives.\u003c/p\u003e\n\u003cp\u003eRecent studies have addressed different aspects of prosthetic control:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\n \u003cp\u003e\u003cstrong\u003eDrishti (2021)\u003c/strong\u003e reviewed trends and challenges in surface EMG for prosthetic applications, emphasizing limitations in portability, feasibility, and cost;\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003e\u003cstrong\u003eTanu (2022)\u003c/strong\u003e investigated SEMG decomposition for hand prosthesis. Despite effectiveness, the approach required multiples electrodes and complex signal processing, limiting real-time performance;\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003e\u003cstrong\u003eForce sensor-based prostheses\u003c/strong\u003e (doi:10.1007/s12647-023-00671-9) improved grasp precision but increased system complexity and cost.\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eUnlike these prior works, the proposed system utilizes a single EEG channel in a SIMO architecture, enabling real-time control of five coordinated finger gestures. It combines low-cost hardware, embedded real-time processing via Arduino Uno, and a 3D-printed mechanical structure, effectively addressing portability, affordability, and simplicity gaps found in previous studies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eKey contributions of this study\u003c/strong\u003e:\u003c/p\u003e\n\u003cp\u003e1. Single EEG channel, instead of multi-channel EEG systems;\u003c/p\u003e\n\u003cp\u003e2. Real-time embedded control using Arduino Uno;\u003c/p\u003e\n\u003cp\u003e3. Full hand articulation with five gestures;\u003c/p\u003e\n\u003cp\u003e4. Low-cost, 3D-printed mechanical design, enabling accessibility and customization.\u003c/p\u003e\n\u003cp\u003eBuilding upon insights from prior research, the following sections detail the proposed methodology, including system architecture, EEG signal acquisition and processing, embedded control, and mechanical hand actuation, highlighting how the SIMO approach addresses previous limitations.\u003c/p\u003e"},{"header":"3. Methodology","content":"\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e3.1 General System Architecture\u003c/h2\u003e \u003cp\u003eThe system follows a SIMO model, where a single EEG input from the prefrontal cortex controls five servomotor outputs corresponding to the fingers of the prosthetic hand. Compared to previous EMG or SEMG based prostheses [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e], this approach offers a low-cost, fully embedded solution with real-time control and full hand articulation.\u003c/p\u003e \u003cp\u003eThe main components include:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eEEG signal acquisition\u003c/b\u003e via a modified AD8232 module;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eArduino Uno\u003c/b\u003e for signal preprocessing, filtering, classification and command generation;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eFive MG996R servomotors\u003c/b\u003e for finger actuation;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003ePassive return mechanism\u003c/b\u003e using springs and fishing wire-pulley transmission;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eA Java -based interface\u003c/b\u003e for real-time monitoring and validation.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the 3D-printed prosthetic hand with mounted MG996R servomotors. The modular architecture ensures precise and synchronized gestures while maintaining simplicity and affordability.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe modular architecture ensures precise, synchronized gestures while maintaining system simplicity and affordability.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e3.2. EEG Signal Acquisition and Processing\u003c/h2\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e3.2.1. Electrode Placement\u003c/h2\u003e \u003cp\u003eEEG signals were acquired using a modified AD8232 module, originally designed for ECG applications, adapted for low-amplitude EEG signals. Electrodes were placed following the International 10\u0026ndash;20 system:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eActive electrode : FP2(right frontal pole);\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eReference electrode : Mastoid(behind the ear);\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eGround electrode: Forearm.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eParticipants were instructed to remain relaxed, minimize head and facial movements, and maintain a neural gaze during data collection to reduce muscular and ocular artifacts. Each acquisition lasted approximately 30 seconds per gesture.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e3.2.2. Hardware modifications and Shielding\u003c/h2\u003e \u003cp\u003eTo improve EEG signal fidelity, the AD8232 module was modified as follows:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e82 Ω resistor: Stabilizes signal amplitude;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e47 \u0026micro;F capacitor: Reduces voltage fluctuations and high-frequency noise;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eElectromagnetic Shielding: The module and connecting wires were enclosed in a grounded metallic case to minimize ambient electrical interference;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eSignal grounding : Reference and ground electrodes connected to a low-impedance ground.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e shows the modified and Shielded AD8232 circuit for EEG acquisition with RC components.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section3\"\u003e \u003ch2\u003e3.2.3. Signal Processing Pipeline (Arduino Uno)\u003c/h2\u003e \u003cp\u003eEEG signals were digitized at 250 Hz using Arduino Uno and processed as follows:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eBand-pass filtering (0.5\u0026ndash;30 Hz) : isolate classical EEG rhythms(delta, theta ,alpha, beta);\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eNotch filter (50/60 Hz) eliminates power-line interference;\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eMoving average filter (5 samples) reduces high-frequency noise;\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eArtifact rejection : segments with amplitude exceeding threshold due to eye or muscle activity were discarded;\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eFeature extraction and classification : filtered signals mapped to predefined gestures using amplitude thresholds and validated against EEG spectral characteristics;\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eDigital command generation: signals control five MG996R servomotors.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eFigures \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e and \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e illustrate the effectiveness of the proposed EEG signal-processing pipeline.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e show the raw EEG signal acquired from the AD8232 module before and filtering, exhibiting significant noise and interference (\u003cb\u003eSNR = -4.74 dB\u003c/b\u003e).\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e presents the same EEG segment after applying the complete processing chain (band-pass, notch, and moving-average filters). The resulting signal shows a clear enhancement in rhythmic components (notably alpha and beta bands), achieving a \u003cb\u003efinal SNR of 24.78dB\u003c/b\u003e, shows effective performance of the filtering and noise-reduction approach.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e \u003ch2\u003e3.2.4. EEG Signal Validation\u003c/h2\u003e \u003cp\u003eTo ensure the reliability of the acquired EEG signals, a spectral analysis was performed using the Fast Fourier Transform (FFT). The objective was to verify whether the recorded signals exhibit characteristics consistent with known neurophysiological EEG rhythms.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e: \u003cb\u003efrequency spectrum of the filtered EEG signal obtained using FFT, showing dominant components in the alpha (8\u0026ndash;13 Hz) and beta (13\u0026ndash;30 Hz) bands\u003c/b\u003e.\u003c/p\u003e \u003cp\u003eThe analysis reveals noticeable peaks within the alpha (8\u0026ndash;13 Hz) and beta (13\u0026ndash;30 Hz) frequency bands. These components are consistent with established EEG patterns associated with relaxed and cognitively active states.\u003c/p\u003e \u003cp\u003eThe presence of these frequency bands confirms that the acquired signals contain physiologically relevant information, despite the use of a low-cost and modified AD8232 acquisition system. Furthermore, the improvement in signal-to-noise ratio (SNR) from \u0026minus;\u0026thinsp;4.74 dB (raw signal) to 24.78 dB (filtered signal) demonstrates the effectiveness of the implemented signal processing pipeline in reducing noise and artifacts.\u003c/p\u003e \u003cp\u003eAlthough the system does not rely on clinical-grade EEG equipment, the results support its feasibility for real-time embedded neuroprosthetic control. Future work will focus on multi-channel acquisition and advanced artifact removal techniques to further improve signal robustness and reproducibility.\u003c/p\u003e \u003cp\u003eThese results should be interpreted as a preliminary validation of EEG signal acquisition using a low-cost setup, highlighting both the potential and limitations of the proposed approach.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.3. Control Strategy\u003c/h2\u003e \u003cp\u003eThe Arduino maps the filtered EEG amplitudes to five gestures states (Open, Close, Zigzag, Gun, Salute) using predefined thresholds. Each gesture corresponding to specific angular positions for the five servomotors, as shown in Table\u0026nbsp;1.\u003c/p\u003e \u003cp\u003e \u003cb\u003eTableau 1 : Servomotor positions corresponding to five classified gestures.\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGesture\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eThumb(\u003csup\u003e0\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIndex(\u003csup\u003e0\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMiddle(\u003csup\u003e0\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eRing(\u003csup\u003e0\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eLittle(\u003csup\u003e0\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOpen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClose\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e175\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e180\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e165\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e170\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e180\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZigzag\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e175\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e180\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e165\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGun\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e165\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e170\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e180\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSalute\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e140\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e160\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e120\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e120\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e160\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\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eAlgorithm Overview\u003c/b\u003e \u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eContinuous monitoring of EEG amplitude;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eGesture activated when amplitude exceeds threshold;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eStepwise angular increments ensures smooth finger transitions.\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eData acquisition and statistical analysis\u003c/b\u003e \u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eEach gesture was repeated 10 times under controlled conditions :(resting state, minimal head movement);\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eEEG was sampled at 250 Hz using Arduino Uno and logged for analysis ;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eDataset :66 samples x 6 variables(1 EEG input\u0026thinsp;+\u0026thinsp;5 fingers outputs);\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTraining set: 90% for model identification;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eValidation set: 10% for performance evaluation.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThis structured dataset enables feasible identification of the dynamic relationship between EEG\u003c/p\u003e \u003cp\u003einputs and finger movements.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.5. Comparative Analysis\u003c/h2\u003e \u003cp\u003eA comparative study was conducted with prior EEG/EMG- based prosthetic systems.\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e2\u003c/span\u003e summarizes the key distinctions.\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 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparison with prior EEG/EMG-based prosthetic hands\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFeature\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMaker 2020\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eDrishti 2021\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTanu 2022\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eProposed work\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEEG control\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePartial\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFull hand articulation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eReal-time embedded control\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePartial\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePartial\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLow-cost/ open-source\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePartial\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePartial\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eData acquisition \u0026amp; analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLimited\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLimited\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLimited\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eFull\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\u003eThis comparison highlights the novelty of the proposed approach: full hand articulation, low-cost embedded control, and comprehensive data acquisition with improved SNR.\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eMechanical design\u003c/b\u003e \u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eTendon-driven actuation : servos pull fishing lines attached to distal phalanges;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eFinger extension: passive via springs mechanism;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePalm structure :houses servos, electronics, and power interface;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eFabrication: PLA (Polylactic Acid) using Fused Deposition Modeling (FDM).\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e illustrates the mechanical design and actuation system of the prosthetic hand, ensuring synchronized finger movements and feasibility while maintaining a low manufacturing cost.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.7. Monitoring Software\u003c/h2\u003e \u003cp\u003eTo monitor the real-time behavior of the EEG signals controlling the prosthetic hand, a dedicated software application was developed in Java using the NetBeans Integrated Development Environment (IDE). This application interfaces directly with the Arduino Uno, which acquires EEG signals from the modified AD8232 sensor placed on the user\u0026rsquo;s forehead.\u003c/p\u003e \u003cp\u003eThe Arduino continuously samples the EEG signals and transmits the data via serial communication (USB) to the Java application running on a computer. The software receives and processes this data in real time, displaying it dynamically on a graphical interface. This setup provides immediate feedback to the user or researcher, enabling assessment of signal quality and verification of prosthetic commands.\u003c/p\u003e \u003cp\u003eMain functionalities of the monitoring software include:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eReal-time signal visualization: EEG voltage values are displayed as a continuous waveform, allowing observation of brain signal features; filtered signals can also be displayed to reduce noise;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eSerial communication handling: The software detects and establishes connection with the correct COM port for uninterrupted data streaming;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThreshold-based gesture recognition (optional): Predefined EEG amplitude thresholds can be visualized to indicate the activation of specific hand gestures;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eCommand translation display: The numerical representation of the EEG signal is shown as corresponding digital commands for the five servomotors, reflecting the mapping of of brain signals into prosthetic finger movements;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eData panel: Numeric EEG and command values are displayed alongside the waveform for calibration, debugging and analysis purposes.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eFigure 8 illustrates the Java-based interface displaying the translation of EEG signals into digital commands for the five fingers. The interface enhances system reliability by providing immediate feedback, enabling real-time verification of gesture execution, and supporting signal quality assessment. By visualizing both EEG activity and the resulting commands, researchers can detect artifacts, assess system responsiveness, and validate the accuracy of prosthetic control.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eFigure 8: Real \u0026ndash;time EEG signal monitoring interface developed in Java using NetBeans.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThis monitoring layer is essential for both system development and experimental validation, ensuring that EEG-derived commands are correctly interpreted and executed by the prosthetic hand with minimal latency.\u003c/p\u003e \u003cp\u003eThe application receives continuous analog EEG data from the AD8232 sensor via Arduino Uno board and displays the signals graphically in real time. Numeric values of the EEG voltage are shown alongside the waveform, enabling effective visualization and analysis of brain activity during prosthetic hand control.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.9. System Identification and SIMO Model in MATLAB\u003c/h2\u003e \u003cp\u003eThe experimental dataset consists of \u003cb\u003e66 samples\u003c/b\u003e with \u003cb\u003e6 variables\u003c/b\u003e each. The first column corresponds to the EEG-derived input command signal, while the remaining five columns represent the angular positions (in degrees) of the prosthetic hand\u0026rsquo;s fingers: thumb, index, middle, ring, and little fingers. The goal is to model the dynamic relationship between the single EEG input and the multiple finger joint outputs, forming a Single-Input Multiple-Output (SIMO) system.\u003c/p\u003e \u003cp\u003eThe dataset was divided into a training set containing \u003cb\u003e90%\u003c/b\u003e of the data for system identification, and a validation set with the remaining \u003cb\u003e10%\u003c/b\u003e for evaluating model performance. This split ensures sufficient data for accurate parameter estimation while providing an unbiased assessment of the models' predictive capabilities. The sampling time was consistent with the EEG signal acquisition frequency.\u003c/p\u003e \u003cp\u003eFive parametric models were tested and compared for each finger:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eTransfer Function (TF);\u003c/b\u003e \u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eAuto-Regressive with eXogenous inputs(ARX);\u003c/b\u003e \u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eAuto-Regressive Moving Average with eXogenous inputs(ARMAX);\u003c/b\u003e \u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eBox-Jenkins(BJ);\u003c/b\u003e \u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eOutput-Error (OE).\u003c/b\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eModel orders were selected based on preliminary analysis and practical considerations. For ARX, ARMAX, and BJ models, the orders were set as follows: \u003cb\u003ena\u0026thinsp;=\u0026thinsp;2\u003c/b\u003e, \u003cb\u003enb\u0026thinsp;=\u0026thinsp;2\u003c/b\u003e, \u003cb\u003enc\u0026thinsp;=\u0026thinsp;2\u003c/b\u003e, \u003cb\u003end\u0026thinsp;=\u0026thinsp;2\u003c/b\u003e, \u003cb\u003enf\u0026thinsp;=\u0026thinsp;2\u003c/b\u003e, with an input-output delay \u003cb\u003enk\u0026thinsp;=\u0026thinsp;1\u003c/b\u003e. The OE model used \u003cb\u003enb\u0026thinsp;=\u0026thinsp;2\u003c/b\u003e, \u003cb\u003enf\u0026thinsp;=\u0026thinsp;2\u003c/b\u003e, and \u003cb\u003enk\u0026thinsp;=\u0026thinsp;1\u003c/b\u003e.\u003c/p\u003e \u003cp\u003eUsing MATLAB\u0026rsquo;s System Identification Toolbox, model parameters were estimated on the training data and their performance validated on the test data. Model accuracy was assessed using \u003cb\u003eRoot Mean Square Error (RMSE)\u003c/b\u003e and \u003cb\u003ecoefficient of determination (R\u003c/b\u003e\u003csup\u003e\u003cb\u003e2\u003c/b\u003e\u003c/sup\u003e\u003cb\u003e)\u003c/b\u003e between predicted outputs and actual finger angles.\u003c/p\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003e3.9.1. Finger-wise Performance\u003c/h2\u003e \u003cp\u003eThe RMSE and R\u003csup\u003e2\u003c/sup\u003e obtained for each finger for the four models are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e3.1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3.1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eRMSE and R\u003csup\u003e2\u003c/sup\u003e values for each finger across different models\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFinger\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eARX\u003c/p\u003e \u003cp\u003eRMSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eARMAX RMSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eBJ RMSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTF RMSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eARX\u003c/p\u003e \u003cp\u003eR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eARMAX R\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eBJ\u003c/p\u003e \u003cp\u003eR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTF\u003c/p\u003e \u003cp\u003eR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThumb\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e14.7152\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e67.6565\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e25.0157\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e64.6890\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.9423\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e-0.2206\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.8331\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e-0.1159\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIndex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.9330\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e50.6500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e23.5132\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e66.5545\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.9961\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.3534\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.8606\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e-0.1165\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e34.3564\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e46.9155\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e42.4491\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e60.9582\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.6459\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.3397\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.4595\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e-0.1147\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e35.3186\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e48.4419\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e37.8849\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e62.8236\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.6475\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.3369\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.5944\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e-0.1153\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLittle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e38.1108\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e106.7037\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e15.6263\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e66.5545\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.6339\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e-1.8698\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.9385\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e-0.1165\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003e3.9.2. Global Performance\u003c/h2\u003e \u003cp\u003eThe evaluate the SIMO system as a whole, RMSE and R\u003csup\u003e2\u003c/sup\u003e were computed over all fingers simultaneously, as shown in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3.2\u003c/span\u003e:\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3.2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eGlobal RMSE and R\u003csup\u003e2\u003c/sup\u003e for the SIMO system across different models\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRMSE(global)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eR\u003csup\u003e2\u003c/sup\u003e(global)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eARX\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e28.6782\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.7784\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eARMAX\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e67.9355\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.2433\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBJ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e30.5287\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.7489\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e64.3527\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.1156\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\u003eThese results indicate that the \u003cb\u003eARX model\u003c/b\u003e provides the best balance between simplicity and accuracy for modeling the prosthetic hand\u0026rsquo;s dynamics.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section3\"\u003e \u003ch2\u003e3.9.3. ANOVA Statistical Analysis\u003c/h2\u003e \u003cp\u003eAn \u003cb\u003eANOVA\u003c/b\u003e test was performed between the EEG input and each finger output to evaluate the statistical significance of the input on the outputs. The resulting p-values are shown in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3.3\u003c/span\u003e:\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3.3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eANOVA p-values between EEG input and finger outputs\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFinger\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSignificance\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThumb\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0101\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStatistically significant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIndex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.2063\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNot statistically significant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMiddle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStatistically significant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStatistically significant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLittle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStatistically significant\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\u003eThe ANOVA results indicate that EEG signals have a statistically significant effect on the movements of the thumb, middle, ring, and little fingers. However, the effect on the index finger is not statistically significant, suggesting weaker or less consistent EEG-based control for this finger compared to the others.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e \u003ch2\u003e3.8.4. Model Selection\u003c/h2\u003e \u003cp\u003eBased on both finger-wise and global metrics, the ARX model was selected as the optimal representation of the prosthetic hand\u0026rsquo;s dynamics. It achieved the \u003cb\u003ehighest R\u003c/b\u003e\u003csup\u003e\u003cb\u003e2\u003c/b\u003e\u003c/sup\u003e and \u003cb\u003elowest RMSE\u003c/b\u003e, while maintaining a simple parametric structure suitable for real-time control applications.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section3\"\u003e \u003ch2\u003e3.8.5. Real vs Predicted Finger Angles\u003c/h2\u003e \u003cp\u003eFigure 8 shows real vs predicted finger angles for the validation dataset.\u003c/p\u003e \u003cp\u003e \u003cb\u003eLegend\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eSolid lines : Real angles;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eDashed lines: Predicted angles;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eColors : Thumb: black/red, Index: green/blue, Middle: cyan/magenta, Ring: yellow/black, Little: red/green.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"4. Experimental Results","content":"\u003cp\u003eThis section presents the experimental validation of the proposed EEG-controlled prosthetic hand prototype. The system was tested in real time to evaluate its ability to accurately reproduce user-intended finger movements based on processed EEG commands. The performance of the ARX system identification model was assessed in terms of accuracy, feasibility, and responsiveness.\u003c/p\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e4.1. Experimental Setup\u003c/h2\u003e \u003cp\u003eThe experimental configuration included:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eProsthetic hand mounted on a test bench, actuated by five MG996R servomotors corresponding to the thumb, index, middle, ring, and little fingers;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eEEG acquisition system based on the AD8232 sensor connected to an Arduino Uno;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eSignal monitoring software developed in Java for real-time visualization of EEG signals;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eControl strategy implementing the ARX model to convert EEG input commands into servo actuation signals;\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThe experimental tests focused on five predefined gestures: opening, closing, salute, zigzag, and pistol sign. Each gesture was performed repeatedly under varying conditions to validate system performance and feasibilty.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e4.2. Gesture Recognition and Validation\u003c/h2\u003e \u003cp\u003eTo quantify gesture success, the recorded EEG command values were mapped to five practical hand gestures as follows:\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4.1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eEEG input ranges and reference finger angles for practical gestures\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEEG input\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGesture\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eReference finger angles [\u003csup\u003e0\u003c/sup\u003e]\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e[1.98;2.27]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOpening\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[0 0 0 0 0]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e]2.27;2.34]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eClosing\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[180\u0026nbsp;180\u0026nbsp;165\u0026nbsp;170\u0026nbsp;180]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e]2.34;2.41]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eZigzag\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[175\u0026nbsp;180\u0026nbsp;165\u0026nbsp;180 0]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e]2.41;2.47]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSalute\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[140\u0026nbsp;160\u0026nbsp;120\u0026nbsp;120\u0026nbsp;160]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e]2.47;5.00]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePistol\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[90 0\u0026nbsp;165\u0026nbsp;170\u0026nbsp;180]\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\u003eRecorded finger angles were compared to these reference angles with a .tolerance of \u0026plusmn;\u0026thinsp;5\u0026deg;. If all finger angles of a sample fell within this range, the gesture was considered successful.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e4.3. Global Success Rate\u003c/h2\u003e \u003cp\u003eBased on the above criteria, the global success rate of gestures was computed using the experimental dataset (EEG.xlsx) containing 66 samples. Each sample included the EEG input and the measured angles of the five fingers.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4.2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSuccess and failure counts of gestures\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResult\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCount\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePercentage\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSucceeded\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e90.91%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFailed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e9.09%\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\u003eThe results show a success rate of \u003cb\u003e90.91%\u003c/b\u003e indicating that the prosthetic hand in real time. This high success rate demonstrates the feasibility of the ARX-based control system.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Synchronization and Real-Time Execution\u003c/h2\u003e \u003cp\u003eA dedicated test was conducted to validate the master\u0026ndash;slave communication protocol:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eMaster system (COM5): generated EEG commands;\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eSlave system (COM6): executed movements on the prosthetic hand;\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThis result confirms the reliability of the communication of link and the capability of the ARX model to ensure stable real-time actuation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003e4.4. Performance Metrics\u003c/h2\u003e \u003cp\u003eTo objectively evaluate the effectiveness of the developed EEG-controlled prosthetic hand, three key performance metrics were assessed:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eMovement precision: This metric quantifies the deviation between the commanded finger angles (desired position) and the executed angles (measured output). A lower deviation indicates higher accuracy of the system in translating EEG commands into mechanical motion.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eResponse time: Defined as the elapsed time between EEG signal acquisition and the initiation of the corresponding finger movement. It reflects the system\u0026rsquo;s ability to operate in real time.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eRepeatability: This evaluates the system\u0026rsquo;s ability to reproduce a given gesture consistently across multiple trials. It is expressed as a percentage of successful gesture reproductions relative to the total attempts.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4.3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003esummarizes the performance results obtained during experimental testing:\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMean Value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStd.Dev.\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrecision error(\u003csup\u003e0\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e9.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026plusmn;\u0026thinsp;31.57\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResponse time(s)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026plusmn;\u0026thinsp;0.00\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRepeatability\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e90.91\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026plusmn;\u0026thinsp;3.54\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\u003eThe results indicate that the prosthetic hand achieves acceptable movement precision with an average angular error of approximately \u003cb\u003e9.94\u0026deg; (\u0026plusmn;\u0026thinsp;31.57).\u003c/b\u003e While this error is higher than the ideal range \u003cb\u003e(\u0026lt;\u0026thinsp;5\u0026deg;)\u003c/b\u003e, it remains within functional limits for basic grasping and gesture reproduction. The response time was measured as negligible \u003cb\u003e(0.00 s)\u003c/b\u003e, reflecting the instantaneous execution of commands once decoded, which is highly favorable for real-time control. Moreover, the system demonstrated strong repeatability \u003cb\u003e(90.91% \u0026plusmn; 3.54)\u003c/b\u003e, confirming its feasibility and reliability in reproducing multiple hand gestures under similar EEG commands.\u003c/p\u003e \u003cp\u003eCollectively, these findings validate the feasibility of the proposed system for real-time prosthetic control, while also highlighting the need for further optimization to reduce angular error and enhance precision.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e4.5. Comparative Analysis with Literature\u003c/h2\u003e \u003cp\u003eWhen compared with related work on surface EMG-based prosthetic control (Drishti, Biomed. Eng. Lett, 2021; Tanu, Measurement, 2022), the proposed EEG-ARX system achieved a comparable gesture recognition success rate (\u0026asymp;\u0026thinsp;90.91%) despite relying on a simpler and low-cost acquisition setup. In contrast to sEMG systems, which typically require multiple electrodes and advanced decomposition algorithms, our EEG-based approach demonstrates feasibility using a single, low-cost sensor with relatively low computational overhead.\u003c/p\u003e \u003cp\u003eIn addition, studies that integrate force sensors for prosthetic hand control (e.g., \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s12647-023-00671-9\u003c/span\u003e\u003cspan address=\"10.1007/s12647-023-00671-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) have reported improved precision and feasibility when combining multimodal feedback. This comparison suggests that integrating EEG with additional sensing modalities, such as force feedback or hybrid EEG-EMG acquisition, could significantly reduce angular errors (mean error\u0026thinsp;\u0026asymp;\u0026thinsp;9.94\u0026deg; in our system) and further enhance reliability.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section2\"\u003e \u003ch2\u003e4.6. Discussion\u003c/h2\u003e \u003cp\u003eThe experimental findings confirm that the EEG-driven prosthetic hand prototype is capable of reliably executing intended hand gestures under real-time conditions. Several key observations can be highlighted:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eThe ARX model effectively maps EEG input signals to motor commands, ensuring stable actuation of finger joints.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThe system achieved an overall success rate of \u003cb\u003e90.91%\u003c/b\u003e, comparable to related studies in non-invasive prosthetic control.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePerformance analysis showed a mean angular error of \u003cb\u003e9.94\u0026deg; (\u0026plusmn;\u0026thinsp;31.57)\u003c/b\u003e, which, although higher than the ideal range \u003cb\u003e(\u0026lt;\u0026thinsp;5\u0026deg;)\u003c/b\u003e, remains within functional limits for basic grasping and gesture reproduction.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eThe response time was negligible \u003cb\u003e(\u0026asymp;\u0026thinsp;0 s)\u003c/b\u003e, confirming the real-time capability of the EEG-based control strategy.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eRepeatability was high \u003cb\u003e(90.91% \u0026plusmn; 3.54)\u003c/b\u003e, highlighting feasibility across multiple trials.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eCompared to sEMG-based approaches, the proposed EEG-ARX system provides a more cost-effective and less invasive solution. However, the larger precision error indicates sensitivity to EEG signal noise and electrode placement. Future research should therefore focus on:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eNoise reduction and adaptive filtering of EEG signals,\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eMultimodal integration (EEG\u0026thinsp;+\u0026thinsp;EMG\u0026thinsp;+\u0026thinsp;force sensors) to enhance accuracy,\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eMechanical refinements of the prosthesis for improved motion smoothness.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eOverall, these results validate the feasibility of using a low-cost EEG-based ARX model for intuitive prosthetic hand control, while also pointing to clear avenues for future improvement.\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eThe experimental results demonstrate the effectiveness of the proposed low-cost EEG-controlled prosthetic hand. The system successfully translated a single EEG input into five distinct finger gestures with minimal latency, validating the proposed Single Input\u0026ndash;Multiple Output (SIMO) control strategy.\u003c/p\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003e5.1. System Performance\u003c/h2\u003e \u003cp\u003e The ARX model provided the best balance between accuracy and simplicity among the tested models. Finger-wise analysis showed high R\u0026sup2; values for the thumb \u003cb\u003e(0.9423\u003c/b\u003e), index \u003cb\u003e(0.9961)\u003c/b\u003e, middle \u003cb\u003e(0.6459)\u003c/b\u003e, ring \u003cb\u003e(0.6475)\u003c/b\u003e, and little fingers \u003cb\u003e(0.6339)\u003c/b\u003e, indicating feasible prediction of angular positions from EEG input. Global system evaluation yielded \u003cb\u003eR\u0026sup2; = 0.7784\u003c/b\u003e, confirming that the SIMO ARX model adequately captured the dynamic relationship between EEG input and finger movements.\u003c/p\u003e \u003cp\u003eThe ANOVA results indicate that EEG signals have a statistically significant effect on the thumb, middle, ring, and little fingers, while the index finger showed non-significant influence. This suggests slightly weaker or less consistent control over the index finger, potentially due to the simplified single-channel EEG acquisition. Nevertheless, the overall gesture execution remains accurate and reliable.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003e5.2. Comparison with Prior Works\u003c/h2\u003e \u003cp\u003eCompared to previous EEG/EMG-based prosthetic systems, the proposed approach offers several advantages:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eFull hand articulation: Unlike prior studies [Drishti, Tanu], which achieved partial finger control, the proposed system enables coordinated movement of all five fingers.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eReal-time embedded control: Using Arduino Uno allows immediate translation of EEG input into servo actuation without external computation.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eLow-cost and accessible design: All components are inexpensive, and the mechanical structure is 3D-printed, making it suitable for low-resource environments.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eComprehensive data acquisition and analysis: EEG and servo data were systematically recorded, enabling system identification, statistical validation, and performance evaluation.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e2\u003c/span\u003e (from Section 3.5) highlights these improvements, demonstrating the novelty and practicality of the proposed system.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec30\" class=\"Section2\"\u003e \u003ch2\u003e5.3. Limitations and Challenges\u003c/h2\u003e \u003cp\u003eDespite these advantages, several limitations should be noted:\u003c/p\u003e \u003cp\u003eSingle EEG channel: While sufficient for simple gestures, complex or highly precise hand movements may require multi-channel EEG acquisition for better accuracy.\u003c/p\u003e \u003cp\u003eIndex finger control: The non-significant ANOVA result suggests that additional preprocessing or signal features may improve control fidelity for the index finger.\u003c/p\u003e \u003cp\u003eNoise and artifacts: EEG signals are inherently noisy, and ambient electrical interference may affect gesture recognition. Hardware modifications and digital filtering partially address this, but further optimization could enhance feasibility.\u003c/p\u003e \u003cp\u003eLimited experimental sample size: With 66 samples, the dataset is relatively small. Future studies could include more subjects and repetitions to improve model generalizability.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec31\" class=\"Section2\"\u003e \u003ch2\u003e5.4. Significance of Findings\u003c/h2\u003e \u003cp\u003eThe study confirms that an affordable, real-time EEG-controlled prosthetic hand can be realized using widely available components. The SIMO ARX model effectively maps brain activity to coordinated finger movements, demonstrating the potential of low-cost neuroprosthetic systems in clinical and low-resource settings.\u003c/p\u003e \u003cp\u003eThese results provide a strong foundation for further research, including:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eMulti-channel EEG integration for enhanced gesture differentiation.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eAdvanced machine-learning algorithms for improved classification and prediction.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eExtended testing with human subjects for real-world performance evaluation.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eIn summary, the proposed system demonstrates that simple, low-cost, and embedded EEG-based control can achieve reliable prosthetic hand operation, bridging the gap between affordability and functionality in neuroprosthetics.\u003c/p\u003e \u003c/div\u003e"},{"header":"6. Conclusion and Future work","content":"\u003cp\u003eThis study presented the design, prototyping, and modeling of a low-cost EEG-controlled prosthetic hand based on a Single Input\u0026ndash;Multiple Output (SIMO) architecture. EEG signals were acquired from the prefrontal cortex using a modified AD8232 module and processed in real time using an Arduino Uno microcontroller. Five MG996R servomotors were used to actuate the fingers, while a 3D-printed mechanical structure with spring-assisted return mechanisms ensured synchronized and repeatable movements.\u003c/p\u003e \u003cdiv id=\"Sec33\" class=\"Section2\"\u003e \u003ch2\u003e6.1. Conclusion\u003c/h2\u003e \u003cp\u003eExperimental results show that the proposed system is capable of translating EEG-derived signals into functional hand gestures with acceptable latency and coordinated finger movements. System identification using the ARX model provided the best performance in terms of accuracy and simplicity, achieving a global coefficient of determination \u003cb\u003e(R\u0026sup2; = 0.7784)\u003c/b\u003e and an acceptable RMSE value.\u003c/p\u003e \u003cp\u003eStatistical analysis (ANOVA) indicated a significant relationship between EEG input signals and most finger outputs, supporting the effectiveness of the single-channel EEG-based SIMO approach.\u003c/p\u003e \u003cp\u003eCompared to existing EEG/EMG-based prosthetic systems, the proposed method offers the following advantages:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eFull hand articulation enabling coordinated multi-finger control\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eReal-time embedded implementation without external computation\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eLow-cost and open-source hardware with 3D-printed mechanical design\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eComprehensive dataset acquisition enabling system identification and validation\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThese results demonstrate the feasibility of low-cost EEG-based prosthetic hand control for embedded neuroprosthetic applications, particularly in resource-constrained environments. However, the system should be interpreted as a proof-of-concept rather than a clinical-grade neuroprosthetic solution.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec34\" class=\"Section2\"\u003e \u003ch2\u003e6.2. Future Work\u003c/h2\u003e \u003cp\u003eAlthough the proposed system shows promising results, several improvements can be explored:\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003e1. Multi-channel EEG acquisition\u003c/h3\u003e\n\u003cp\u003eAdding additional EEG channels may improve signal robustness and enhance control performance, particularly for fingers with weaker correlation such as the index finger.\u003c/p\u003e\n\u003ch3\u003e2. Advanced signal processing and machine learning methods\u003c/h3\u003e\n\u003cp\u003eTechniques such as adaptive filtering, feature extraction, and classifiers (e.g., SVM or neural networks) could improve gesture classification accuracy.\u003c/p\u003e\n\u003ch3\u003e3. Extended experimental validation\u003c/h3\u003e\n\u003cp\u003eFuture studies should include larger datasets and multiple participants to evaluate generalization, robustness, and real-world usability.\u003c/p\u003e\n\u003ch3\u003e4. Mechanical system optimization\u003c/h3\u003e\n\u003cp\u003eImprovements in tendon routing and spring mechanisms could enhance movement smoothness, durability, and mechanical efficiency.\u003c/p\u003e\n\u003ch3\u003e5. Portable and autonomous power integration\u003c/h3\u003e\n\u003cp\u003eThe integration of compact energy storage systems would enable fully wearable operation.\u003c/p\u003e\n\u003ch3\u003e6. Integration with assistive and rehabilitation systems\u003c/h3\u003e\n\u003cp\u003eThe prosthetic platform could be extended to other BCI or assistive technologies for rehabilitation and human\u0026ndash;machine interaction applications.\u003c/p\u003e \u003cp\u003eThis work demonstrates a practical and low-cost approach to EEG-based prosthetic hand control, combining embedded systems, signal processing, and mechanical design. It provides a foundation for further development of accessible neuroprosthetic devices that balance affordability, functionality, and real-time performance.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003e \u003cb\u003eConflicts of interest\u003c/b\u003e:\u003c/h2\u003e \u003cp\u003eThe authors declare that they have no conflict of interest.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eEthical approval:\u003c/strong\u003e \u003cp\u003eThe study wss approved by the institutional Ethics Committee for Research of Human Health of the University of Douala (Authorization N\u003csup\u003e0\u003c/sup\u003e4731CEI-Udo/01/2025/T). All experimental procedures were performed in accordance with relevant guidelines and regulations.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eConsent to Participate:\u003c/strong\u003e \u003cp\u003eThe author, who served as the sole participant in this study, provided informed consent to participate.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding:\u003c/h2\u003e \u003cp\u003eThis research received no external funding.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAuthor Contributions StatementJAWNM conceived the research idea, designed the prosthetic hand system, implemented the EEG acquisition setup, and developed the Arduino-based control architecture. JAMW also performed the signal processing, system modeling, and experimental validation, including FFT analysis and performance evaluation of the SIMO system.LNN supervised the research work, contributed to the methodological guidance, provided critical feedback on system design and data interpretation, and reviewed the manuscript for scientific quality and structure.All authors reviewed and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe authors would like to thank the Laboratory of Computer Science Engineering and Automation, University of Douala, for providing technical support and facilities for this research.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eYes. I used or generated research data in the study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eDrishti D. Recent trends and challenges of surface electromyography in prosthetic applications. Biomed Eng Lett, 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTanu. Decomposition and evaluation of SEMG for hand prostheses control in measurement. Measurement, 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi H, Xu Z, Zhou J. Force sensor applications in prosthetic hands: A review. Biomedizinische Technik/Biomedical Eng. 2023;68(1):45\u0026ndash;58. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s12647-023-00671-9\u003c/span\u003e\u003cspan address=\"10.1007/s12647-023-00671-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLebedev M, Nicolelis M. Brain\u0026ndash;machine interfaces: past, present and future. Trends Neurosci. 2006;29(9):536\u0026ndash;46.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNicolas-Alonso LF, Gomez-Gil J. Brain Computer Interfaces, a Review. Sensors. 2012;12(2):1211\u0026ndash;79.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMishra AK, Rao AS, Jha PR. Low-cost prosthetic hand with EEG-based control for disabled persons. Procedia Comput Sci. 2018;133:85\u0026ndash;92.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarim KS, Rahman MS, Rahman MT. Design and implementation of prosthetic hand using Arduino and EEG sensor. Int J Sci Eng Res. 2018;9(3):317\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSiahaan MA, Tarigan RK, Siregar AH. Design of 3D printed prosthetic hand actuated by Arduino-based myoelectric signals. IOP Conf Series: Mater Sci Eng, 546, 2019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBernard S, Fischer L, Maier MH. Real-time control of an upper-limb neuroprosthesis based on non-invasive EEG. Biomed Signal Process Control, 87, 2024.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen Y, Huang J. Comparative Analysis of System Identification Techniques for Prosthetic Finger Control. IEEE Access. 2024;12:50711\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBello AT, Singh K, Jain R. EEG signal acquisition and gesture recognition for low-cost neuroprosthetics. J Neural Eng, 21, 1, 2024.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSmith LO, Thompson K. Performance evaluation of OE and ARX models in bio-inspired control. Int J Control Autom Syst. 2024;23:421\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJohn TU, George MP, Iyer RS. Design and evaluation of a neuroprosthetic hand controlled via EEG signals on Arduino platform. Int J Adv Rob Syst, 22, 2, 2024.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMbemba R, Diallo AK, Nguema FB. Modelling and control of a 5-DOF prosthetic hand using ARX and neural networks. Front Rob AI, 11, Article 2201432, 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdewale JN, Yusuf M, Bamidele KO. 3D printed smart prosthetic hand using embedded Arduino and biosensors. Sens Actuators A: Phys, 367, 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang H, Wu L, Li Y. Integration of force and EEG feedback in human-machine interfaces: Application to prosthetics. IEEE Trans Biomed Circuits Syst. 2025;18(2):304\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDjamal FN, Hidayat AR, Wibowo S. EEG-based Human\u0026ndash;Machine Interface for prosthetic finger control using ARMAX models. Biomed Eng Lett. 2025;15:207\u0026ndash;15.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eElenga MR, Nguema P. Comparison of system identification models for real-time prosthetic finger actuation. Control Eng Pract, 140, 2025.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"discover-applied-sciences","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Discover Applied Sciences](https://link.springer.com/journal/42452)","snPcode":"42452","submissionUrl":"https://submission.springernature.com/new-submission/42452/3","title":"Discover Applied Sciences","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Electroencephalography (EEG), Brain-Computer Interface(BCI), Prosthetic hand, Embedded systems, Signal Processing, SIMO Modeling, System Identification, Real-Time","lastPublishedDoi":"10.21203/rs.3.rs-9413097/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9413097/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study presents the design and experimental validation of a low-cost EEG-controlled prosthetic hand capable of executing multiple hand gestures using a single-channel input. EEG signals were acquired from the prefrontal region using a modified AD8232 module and processed in real time using an Arduino Uno microcontroller to control five MG996R servomotors corresponding to five predefined gestures (open, close, greeting, pistol, and zigzag). To address the challenges associated with low-cost EEG acquisition, a complete signal processing pipeline was implemented, including band-pass filtering (0.5\u0026ndash;30 Hz), notch filtering (50 Hz), and moving-average smoothing. In addition, spectral validation was performed using Fast Fourier Transform (FFT) analysis. The results demonstrate the presence of dominant frequency components in the alpha (8\u0026ndash;13 Hz) and beta (13\u0026ndash;30 Hz) bands, suggesting the presence of physiologically relevant EEG signals despite hardware limitations. The prosthetic hand was fabricated using 3D printing (PLA) and incorporates a cable-driven actuation system with a passive spring return mechanism. A Java-based interface was developed for real-time monitoring of EEG signals and finger movements. System identification techniques, including ARX, ARMAX, Box-Jenkins, and Transfer Function models, were used to establish the relationship between EEG input and finger motion outputs in a SIMO framework. Among these, the ARX model achieved the best performance (R\u0026sup2; = 0.7784, RMSE\u0026thinsp;=\u0026thinsp;28.68\u0026deg;).\u003c/p\u003e \u003cp\u003eWhile the results demonstrate the feasibility of low-cost EEG-based prosthetic control, the study highlights the importance of signal validation and noise mitigation in such systems. The proposed approach provides a promising and accessible solution for neuroprosthetic applications in low-resource environments.\u003c/p\u003e","manuscriptTitle":"Low-Cost EEG-Based Prosthetic Hand Control: Design, Real-Time Implementation, and SIMO Modeling","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-18 10:52:51","doi":"10.21203/rs.3.rs-9413097/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-23T18:28:49+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"247465251617420324558370920610741097720","date":"2026-05-16T07:15:40+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"58898868001845954402991721299480483425","date":"2026-05-09T14:13:09+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-05-07T06:22:22+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-23T11:21:34+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-17T08:57:59+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-17T08:57:23+00:00","index":"","fulltext":""},{"type":"submitted","content":"Discover Applied Sciences","date":"2026-04-14T09:02:46+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"discover-applied-sciences","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Discover Applied Sciences](https://link.springer.com/journal/42452)","snPcode":"42452","submissionUrl":"https://submission.springernature.com/new-submission/42452/3","title":"Discover Applied Sciences","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"42896988-276e-4b35-92c3-e2c84db9bbe1","owner":[],"postedDate":"May 18th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-23T18:28:49+00:00","index":35,"fulltext":""},{"type":"reviewerAgreed","content":"247465251617420324558370920610741097720","date":"2026-05-16T07:15:40+00:00","index":32,"fulltext":""},{"type":"reviewerAgreed","content":"58898868001845954402991721299480483425","date":"2026-05-09T14:13:09+00:00","index":21,"fulltext":""},{"type":"reviewersInvited","content":"12","date":"2026-05-07T06:22:22+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-18T10:52:51+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-18 10:52:51","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9413097","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9413097","identity":"rs-9413097","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}