Page 56 - IJAMD-2-3

International Journal of AI for
Materials and Design Intelligent interactive textile in healthcare

Alternative gesture-tracking models comparable to popular enhancing user confidence. Future research could further
54
hand landmark solutions include OpenPose, Leap Motion, explore these user-centered design approaches to ensure
and other proprietary landmark detection frameworks that gesture recognition technologies feel neither intrusive nor
similarly leverage neural networks for real-time gesture burdensome.
tracking. 63,64
3. Methodology
Raw data from e-textiles, such as pressure signals,
optical signals, or electromyography, benefit from robust 3.1. Interviews
classification algorithms capable of interpreting complex This research adopted a multi-phase participatory
spatiotemporal patterns. Neural architectures, such design approach over a 37-month period. The study
5,59
as convolutional neural networks or attention-based commenced with a workshop (co-design session A;
transformers, interpret subtle gesture variations and adapt December 22, 2021) at SKH Calvary Church in Wong Tin
to user-specific differences. Guo et al., demonstrated that a Sin, establishing the foundation for community-driven
61
one-dimensional convolutional neural network trained on design principles. Subsequent semi-structured interviews
sEMG signals could classify 10 distinct hand poses relevant (co-design session B) with the research lead and the OT
to stroke therapy, achieving accuracy exceeding 90%. Such lead (February 7, 2023) at the Core Centre at WTSDHC
data-driven modeling is critical for real-time feedback, provided institutional perspectives on implementation
flagging suboptimal movements or guiding corrective steps requirements and healthcare objectives. User experience
during rehabilitation. Nevertheless, deep learning’s power evaluation (co-design session C) was conducted through
consumption and computational overhead pose challenges structured interviews with co-designers and end-users
for embedded textile platforms with limited battery or (March 25, 2024) at the same place. The study received
hardware resources. Researchers are thus exploring ethical approval from the Institutional Review Board at
5
lightweight or optimized neural models deployable on the authors’ affiliated university. All participants provided
microcontrollers. Edge computing can minimize reliance informed consent before the commencement of the
on cloud connectivity, which is advantageous for remote or interviews.
resource-limited clinical environments. Prototypes also
60
employ data fusion techniques, combining optical gesture This methodological sequence allowed for iterative
tracking with sEMG signals, refining gesture accuracy development while maintaining alignment with both
community needs and healthcare requirements. The
without significantly increasing hardware complexity. 61
technical development phase implemented these insights
2.3.3. Healthcare applications and user acceptance through a gesture recognition illuminative knitted
textile system, comprising three interactive interfaces.
Gesture-driven textiles hold substantial promise in Each interface was developed using specific interaction
healthcare, including telemedicine, physical therapy, and modalities: hand gesture recognition, shoulder movement
elderly care. 54,62 A textile-based gesture interface could detection, and head movement recognition, with
enable older adults to call for help using simple hand signs corresponding visual feedback mechanisms. The final
rather than navigating small buttons. In stroke recovery, evaluation phase employed qualitative user feedback,
interactive textiles or gloves could monitor progress enabling systematic refinement of the system’s technical
during range-of-motion exercises, providing real-time parameters while maintaining therapeutic efficacy and
feedback and gamified incentives. Practical deployment, user engagement objectives.
however, demands attention to usability, washability, and
robustness. E-textiles integrated into hospital curtains or 3.1.1. Co-design workshop
5,6
seat covers must withstand repeated cleaning cycles, and This study adopted a systematic participatory action
wearable gloves must maintain accurate sensor functions research framework, comprising three distinct yet
despite mechanical stresses.
interconnected phases. The initial phase employed a
User acceptance among older adults remains critical. participatory design methodology to engage members of
Oudah et al., noted seniors’ skepticism toward unfamiliar the WTSDHC in a 3-h co-design workshop (co-design
62
technologies as a potential barrier to adoption. Co-design session A). A total of 11 participants, including five staff
strategies involving healthcare staff, caregivers, and older members and OTs and six center members, took part in
patients can yield intuitive gestures and esthetically the session. The workshop was divided into two parts.
pleasing fabrics. Studies like Tan et al. emphasize visual The first part consisted of a presentation introducing
54
feedback – illuminative textiles can confirm correct the concept and applications of intelligent illuminative
gesture detection by changing color or brightness, textiles, aiming to provide participants with foundational

Volume 2 Issue 3 (2025) 50 doi: 10.36922/IJAMD025170013