Human body motion sensors are crucial for wearable electronics and intelligent medicine, requiring ultralight weight, flexibility, biocompatibility, and high precision. Conventional sensors struggle to meet these criteria due to rigid structures, high-density materials, and complex assembly. Existing skin-interface devices, like E-skins and textile sensors, offer potential but face challenges in long-term wear comfort, biocompatibility, and signal classification. Moisture and gas permeability are critical for comfort and biocompatibility, while signal classification often requires sophisticated techniques like machine learning to handle subtle movements. This study aims to overcome these limitations by introducing a novel AFMS based on RAPSF, which addresses the need for a lightweight, flexible, biocompatible, and highly accurate motion sensor.

The introduction reviews existing technologies for human body motion monitoring, including sensors based on liquid metals, hydrogels, graphene aerogels, functional fibers, and piezoelectric materials. The limitations of these technologies are highlighted, focusing on the challenges of achieving optimal moisture/gas permeability, biocompatibility, and accurate signal classification, particularly for subtle movements. The authors emphasize the increasing importance of integrating artificial intelligence (AI), specifically machine learning, for improved signal classification and the development of next-generation healthcare systems.

The RAPSFs were fabricated using a modified blow-spinning system with environmental control. This involved a two-channel six-needle module, a compressed air temperature regulator, and a temperature-controlled fiber collector followed by UV radiation. The process utilizes phase separation during jet flow by controlling air temperature and flow rate, and unilateral pre-nucleation by controlling collector temperature. Different concentrations of dichloromethane were explored to optimize the porous structure. The fabricated RAPSFs were characterized using SEM, XRD, XPS, BET, and other techniques to analyze their micromorphological characteristics and electrical properties. The AFMS was then fabricated by layer-by-layer assembly of RAPSFs, a PAN substrate, AgNWs electrodes, and PVB adhesive. Its breathability, moisture permeability, and biocompatibility were assessed through gas permeability tests, moisture permeability measurements, and in vivo biocompatibility testing on mice. The AFMS's motion tracking capabilities were demonstrated by monitoring finger, elbow, knee, and throat movements. The collected signals were processed using DFT analysis for simple motions and machine learning (XGBoost, KNN, SVM) for complex throat motions (coughing, swallowing, speech).

The RAPSF exhibited a porous and radial anisotropic structure, with adjustable resistance (80-213 Ω) depending on the bending radius. The AFMS achieved an ultra-low density (68.7 mg cm⁻³), high moisture permeability (59.7 g m⁻² h⁻¹), and high gas permeability (low gas resistance of 9.8 × 10⁻³ Pa). In vivo biocompatibility tests showed minimal inflammation and no irreversible tissue necrosis. The AFMS demonstrated effective motion tracking of fingers, elbows, and knees through DFT analysis. Machine learning, specifically XGBoost, achieved over 85% accuracy in classifying throat movements (coughing, swallowing, and speech), showing promise for early detection of viral throat illnesses.

The successful fabrication of the ultralight, flexible, biocompatible AFMS addresses several critical limitations of existing motion sensors. The combination of RAPSF's unique properties and AI-based signal processing enables highly accurate and nuanced motion detection. The high accuracy in throat motion classification demonstrates the potential for early diagnosis of respiratory illnesses. The findings highlight the potential of this technology for various applications in wearable electronics and next-generation intelligent healthcare systems.

This study successfully demonstrated a novel ultralight, flexible, and biocompatible all-fiber motion sensor. The sensor's unique properties, combined with machine learning algorithms, achieve high accuracy in motion and throat condition recognition. This work contributes significantly to the advancement of wearable AI healthcare technology and opens up opportunities for future research in more complex motion tracking and disease monitoring applications.

While the study demonstrates excellent performance in several key areas, future studies could explore the long-term stability of the sensor under various environmental conditions and diverse physiological situations. Further research is also needed to refine the machine learning models for more robust and generalizable performance across different individuals and contexts. The sample size in the biocompatibility study was relatively small and a larger-scale study would strengthen these results.