The micro-nanoscale devices’ design offers significant advantages, including precise control, fast response, high energy-density efficiency and cost-effective production. These benefits have driven substantial advances in micro- and nano-electromechanical systems (MEMS and NEMS), leading to the development of different materials and manufacturing techniques that support a wide range of MEMS devices (Welburn et al., 2025). Pressure sensors, accelerometers, gyroscopes and temperature sensors provide essential feedback for fine control and precise handling, while MEMS resonators contribute to frequency stability, efficient signal processing and energy harvesting, increasing the overall performance and autonomy of the prosthetic system. The signals, collected by the sensors, are processed and converted into coordinated movements through miniaturized actuators and integrated mechatronic systems, enabling the hand to perform complex operations, i.e., gripping, rotation and dynamic adaptation to surfaces of varying shapes and textures.

This precise mechanical control, obtained thanks to the synergistic interaction between the MEMS components and the prosthetic structure, is essential to faithfully reproduce the functional dynamics of the human hand, ensuring fluid kinematics and fine dexterity. Recent studies have shown significant advances in the integration of MEMS into prosthetic hands, as well as the application of Machine Learning (ML) methods based on neural networks. The study (Ke et al., 2022) introduced a novel thumb structure and integrated fingertip tactile sensor to enable coupled movement and precision manipulation. The prosthetic hand’s geometry was optimized using a new metric and Monte Carlo method, resulting in over 10% improved grasping ability and successful human-like thumb opposition in experiments. For amputees using prostheses without sensory feedback, the biggest challenge is handling unexpected events, especially object slippage during grasp and manipulation tasks. The study (Gentile et al., 2020) introduced a low-cost, low-computational touch-and-slippage detection algorithm for prosthetic hands, relying solely on the normal force component from monoaxial sensors. Results showed a 99.4% success rate in detecting slippage events with a computation time of just 4.9 ms, minimizing false positives/negatives. From materials side, ultra-high molecular weight polyethylene (UHMWPE) is widely used in joint replacements, but its wear debris could cause complications like aseptic loosening. The study (Dalli et al., 2023) evaluated a novel hip prosthesis with a unidirectional cylindrical design, aiming to reduce wear compared to the traditional ball-and-socket design. The study highlighted the importance of considering material behavior like molecular orientation in wear models, especially for innovative prosthesis designs. This insight could improve the durability and performance of future joint replacements. A review (Zhang, et al., 2025) analyzed sEMG-IMU sensor fusion techniques for upper limb movement pattern recognition, breaking down the signal generation of surface electromyography (sEMG) and inertial measurement units (IMU). It examined multisensory fusion strategies to improve accuracy and reliability in movements.

Advanced sensor data and machine learning algorithms, such as Stacked Bidirectional Long Short-Term Memory (SBLSTM), Adaptive Neuro-Fuzzy Inference System (ANFIS), Convolutional Neural Network (CNN), Logistic Regression (LR), and K-Nearest Neighbours (KNN), were compared, with the SBLSTM model achieving the highest accuracy (96.3%) and fastest inference time (25 ms), making it most suitable for real-time prosthetic control (Kumar and Pratihar, 2025). Another study (Yang et al., 2025) developed a lightweight prosthetic hand with 19 degrees of freedom, using shape-memory alloy actuators to precisely control the fingers and wrist, thus approaching the functionality of the human hand (Sankar et al., 2025). introduced a biomimetic prosthetic hand with soft robotic joints and a rigid exoskeleton, integrating three independent layers of neuromorphic touch sensors for advanced sensory perception. Recent advances in computer-aided design, engineering and manufacturing (CAD-CAE-CAM) have profoundly transformed the design and production of medical devices, offering opportunities for customization, precision and rapid prototyping. In the orthopaedic sector, as in dental one, CAD-based workflows and 3D printing technologies are continuously evolving to reduce treatment procedures, enhance material efficiency and improve patient outcomes. In particular, additive manufacturing plays a pivotal role with its rapid expansion across diverse fields, i.e., maritime, aerospace and civil engineering, that has drastically shortened production times for new products. This approach enables the creation of complex three-dimensional geometries directly from digital models generated within a CAD environment, facilitating accurate and repeatable fabrication processes (Baiamonte et al., 2024). In the field of prosthetic hand design, this evolution is closely linked to the adoption of additive manufacturing, which marks a shift away from traditional subtractive techniques like milling. Additive technologies enable the creation of customized, anatomically accurate medical devices, ultimately improving patient outcomes and restoring the ability to perform daily activities with near-physiological interaction with the environment. Conversely, conventional precision manufacturing remains essential when achieving specific surface roughness and dimensional accuracy is critical, particularly in high-precision mechanical applications where both economic efficiency and product quality must be ensured (Calì et al., 2025). Furthermore, finite element analysis (FEA) is increasingly employed to evaluate the mechanical resistance of components under load, ensuring compliance with defined design and safety requirements (Seminara et al., 2022). Within this scoping review, the current state of prosthetic hand development is analysed, comparing various CAD modeling technologies and highlighting critical mechanical and design aspects. The purpose is to systematically map and evaluate current research on the mechanical modelling approaches and the integration of MEMS sensors with EMG signals in prosthetic hand design, with the aim of identifying their contribution to dynamic control, biomechanical performance and user-centred functionality. Aspects of microsurgery and interface physiology were also considered, including the use of MEMS as electrodes for nerve interaction, common issues with implantable prostheses such as potential immune reactions to materials, and the importance of targeted microsurgical debridement and bone stump curettage, removing avascular or necrotic areas prior to prosthetic implantation. The findings emerging from the analysed literature show that MEMS-based sensing systems significantly enhance grasping force modulation, motion stability and real-time control responsiveness, while advanced mechanical architectures, such as tendon-driven mechanisms, multi-link transmissions and synergy-inspired designs, enable more natural, coordinated and adaptable movements. By consolidating these technological trends, the review provides an updated overview of the most effective strategies currently used to approach human-like dexterity, highlighting both the progress achieved and the challenges that still limit the full biomimetic integration of prosthetic hands. The research study is divided into four sections, excluding the introduction: Section 2 shows the materials and methods adopted, Section 3 draws the results and discussion of the scoping review, Section 4 discusses the microsurgery and physiological aspects of interface systems and finally Section 5 describes the conclusions.

This scoping literature review was conducted following the PRISMA guidelines (Moher et al., 2009). For all studies, the authors performed an electronic search, conducted to identify articles that address the adopted techniques for MEMS sensors and mechanical modelling of prosthetic hands; the authors manually screened reference lists of included articles. The approach adopted can be considered inclusive, as a wide range of studies were considered, identifying gaps in research and without applying overly restrictive selection criteria. This has allowed authors to obtain a more complete picture of what has already been studied and to identify any areas that are still unexplored, as well as providing a basis for more detailed future studies. The following PRISMA flow diagram (Figure 1) describes the literature screening process adopted in the scoping review and inclusion process (Figure 2).

A total of 115 articles were initially identified through the database search and 109 remained after duplicate removal. Following a detailed screening process and full-text assessment, 7 studies met the inclusion criteria and were included in this scoping review. Table 2 summarizes the main technical and mechanical characteristics of the analyzed prosthetic hand systems, including mechanism placement, wearable weight, force capabilities, and clinical applications. The following sections provide a more in-depth comparative discussion of these technologies, focusing on the key findings and highlighting how each approach contributes to improve dexterity, adaptability and sensory feedback in modern prosthetic hands.

As early as the beginning of the 20th century, Roy envisioned scenarios in which neurosurgery could enable functional restoration or even augmentation of body parts through the implantation of small electronic devices, then referred to as “smart systems” (Roy et al., 2001). MEMS are electrodes that can work as sensors or actuators, and they played a key role in the development of so-called interface systems, so points of interaction with a specific anatomical structure (i.e., brain, muscle, nerve) (Guo et al., 2022; HajjHassan et al., 2008). Indeed, the name of these systems changes depending on the structure with which the electrode interfaces, therefore we mainly distinguish between muscle-interface and neural-interface systems. In myoelectric interfaces, signals are derived from electromyographic recordings of agonist-antagonist muscle groups in the limb stump, obtained via surface electrodes applied to the skin or through implantable electrodes (Tapia et al., 2020; Chen et al., 2023). Among the latter, epimysial electrodes are sutured onto the connective tissue sheath, while intramuscular electrodes penetrate the epimysium (Earley et al., 2025; Yildiz et al., 2020).

In neural-interface systems, the main structures interacting with electrodes are the brain or the peripheral nerve (Russell et al., 2019). Brain-computer interfaces rely on electro-encephaloghraphy (EEG) recordings, in which cortical waves represent inputs to generate signals translated into motion, and the interaction electrodes may be surface electrodes placed by a scalp helmet (Orban et al., 2022) or subdural electrodes implanted directly on the motor cortex (Mokienko 2024). Electrodes that interact with the nerve can be classified as either extra-neural or intraneural. The former varies in shape and can be sutured onto the epineurium or arranged around the nerve (helical, spiral cuff) (Yildiz et al., 2020; Larson and Meng, 2020). Implantation of intraneural electrodes requires exposure of nerve fascicles after an epineurium incision, taking care not to compromise the microcirculation (vasa nervorum). The electrode is then inserted perpendicularly, trying to involve as many fascicles as possible (Bumbaširević et al., 2019). Radial and ulnar nerves are most commonly used in intraneural microsurgery, as they have both sensory and motor components (Yildiz et al., 2020; Larson and Meng, 2020). The electrode records depolarisation propagating in a “saltatory way” along the Ranvier nodes, where voltage-dependent Na⁺ channels alternate along the myelinated motor fibers (Stadelmann et al., 2019). It may also record the resulting striated muscle contraction electromyographically, which occurs once the action potential has reached the neuro-muscular junction (Chen et al., 2023). As evinced from recent studies, targeted electrical stimulation may modify muscle physiology inducing muscle adaptation thanks to an optimal recruitment of fast-twitch fibers (Lin et al., 2025). The development of neuromas at the distal end of damaged nerve is a post-amputation phenomenon that can occur later. To reduce this risk, some techniques to stimulate axonal sprouting have been developed, such as the regenerative peripheral nerve interface (RPNI) and targeted muscle reinnervation (TMR). In RPNI, the free end of the nerve is implanted into free devascularised muscle clusters, after microsurgical neuroma excision if necessary (Leach et al., 2023). TMR involves grafting the main upstream nerves (radial, ulnar, median and muscolocutaneous nerves) into the pectoralis muscles. Once the nerve signal could activate pectoral muscles contraction, it is recorded by chest electrodes and translated into prosthetic movement (Mauch et al., 2023). A common challenge for all implantable prostheses is the risk of a hyperactive immune response, in which the body recognizes the implant as foreign, potentially leading to rejection (Mariani et al., 2019). Bioengineering advances in recent decades have encouraged the study and development of biocompatible materials that do not hinder bio-integration thanks to their very low immunogenicity, but rather accompany the linear growth of tissues (Ciulla et al., 2023). Although peripheral Schwann cells can replicate and stimulate nerve regeneration, a phenomenon known as Wallerian degeneration may occur after amputation. In these cases, if the stump nerve is not properly stimulated and is subjected to mechanical or inflammatory stress, it can degenerate, hindering the engraftment of sutures (Tian et al., 2024). Some malignant tumours, such as hand sarcomas, may destroy adjacent tissues prior to resection, compromising the integrity of vessels and nerves. In these cases, adjuvant radiotherapy could also slow down the phases of osseointegration and peripheral myelin regeneration, leading to the development of excessive fibrotic scar tissue around neurovascular structures and making microsurgery particularly challenging (Zagardo et al., 2025). In such cases, a targeted microsurgical debridement and curettage of the bone stump, removing avascular or necrotic areas, should be performed before prosthetic implantation. Secondary conditions following limb loss, such as phantom limb syndrome (PLS), involve maladaptive neuroplasticity and may hinder functional integration of the prosthesis due to excessive pain (Kaur and Guan, 2018). However, when prosthetic implantation is combined with targeted physiokinesitherapy, it may promote adaptive neuroplasticity in cerebral area of hand control, counteracting PLS development (Makin and Flor, 2020). Ultimately, a multidisciplinary collaboration between neurosurgeons and biomedical engineers in the refinement of these devices will be essential to optimize treatment strategies, improve quality of life, and support the patient’s psychological well-being (Marrone et al., 2025).

The upper limbs are essential not only for performing daily tasks such as grasping, holding and manipulating objects, but also for enabling social and communicative interactions, including gestures and handshakes. Consequently, for individuals experiencing upper-limb amputation, the sudden loss of motor functionality results in physical and socio-economic consequences, severely reducing independence and overall quality of life. This scoping review on MEMS sensors and mechanical design for prosthetic hand control highlights how innovations in mechanical design, actuation systems and materials engineering are progressively narrowing the performance gap between artificial and biological hands. The prosthetic prototypes examined range from highly articulated, sensor-rich hands featuring MEMS-based control, to simplified tendon- or linkage-driven mechanisms, each pursuing an optimal balance between dexterity, compactness and mechanical robustness. Relocating actuators in the forearm to reduce distal inertia, incorporating compliant mechanisms for adaptive grasping and optimizing gear ratios, stiffness, and damping parameters, contribute to achieve human-like motion dynamics under realistic loading conditions.

Regarding the first research question, results clearly demonstrate that MEMS sensors significantly enhance prosthetic hand performance by enabling accurate and responsive grasp force feedback. Their high-resolution pressure, inertial and thermal measurements allow fine modulation of gripping forces, improve stability during dynamic manipulation and support robust real-time control loops that closely approximate physiological sensorimotor pathways. Thus, interaction with objects of varying stiffness and geometry is improved. Regarding the second research question, the analysis shows that a wide range of mechanical modelling approaches, including tendon-driven systems, differential mechanisms, multi-link kinematic chains, flexure-based joints and synergy-inspired concepts, are employed to reproduce the complexity of human hand motion. Supported by CAD parametric modelling, finite element analysis (FEA) and dynamic simulation, these models enable refinement of joint biomechanics, improvements in transmission efficiency and in motion patterns, closely mirroring anatomical function.

Despite substantial advancements, including lighter structures, improved stability and increased control precision several challenges remain. These include achieving higher force-to-weight ratios, ensuring long-term structural durability and improve the adaptability of the fingers, minimizing false positives from the EMG sensors. Ultimately, the future of prosthetic hand development lies in advanced materials and neural interfacing, incorporating myoelectric or peripheral nerve interfaces that directly connect artificial limbs with the nervous system. This approach will develop devices that combine structural robustness with the natural adaptability and perceptual feedback characteristic of the human hand.