Abstract
<title>Abstract</title> <p> <bold>Background</bold> As population ageing accelerates, pressure injury (PI) management among community-dwelling older adults has become an urgent challenge in transitional and continuing care. Older adults with PIs often have limited mobility, impaired nutritional status, sensory decline, and multiple chronic comorbidities, all of which increase the risks of wound progression, infection, delayed healing, repeated healthcare visits, and reduced quality of life. In home-based settings, PI care relies heavily on family caregivers. Although they are usually non-professionals, they are expected to undertake tasks such as skin observation, pressure relief, dressing-related care, nutritional support, and timely healthcare-seeking decision-making. However, family caregivers often lack standardized assessment skills and the ability to recognize early wound deterioration. Although existing Internet Plus Nursing services have improved the accessibility of out-of-hospital nursing care to some extent, they are still largely based on single consultations or one-time home visits and have not fully addressed fragmented assessment, discontinuous follow-up, and insufficient decision support in out-of-hospital PI management. These limitations may delay early identification, timely care adjustment, and referral. This study aimed to evaluate the effects of a digital home care pathway supported by a YOLOv8s-assisted PI staging tool on wound healing outcomes across different PI stages, family caregiver burden, and PI-related service costs among community-dwelling older adults. <bold>Objective</bold> This study aimed to evaluate the effects of a YOLOv8s-assisted WeChat mini-program integrated into an Internet Plus Nursing pathway on caregiver burden, PI-related service costs, and wound healing outcomes among community-dwelling older adults with stage 1–4 PIs. <bold>Methods</bold> This was a prospective, single-center, non-randomized, parallel-group controlled study conducted at The First Hospital of Jiaxing, Zhejiang Province, China, from January 2025 to May 2026. Using convenience sampling, 120 caregiver–patient dyads involving community-dwelling older adults with stage 1–4 PIs and their family caregivers were consecutively enrolled and assigned to the intervention or control group by alternate allocation based on admission sequence. The control group received Internet Plus Nursing care(IPN group). In addition to standard care, the intervention group used the “YOLOv8s-PI Assistant” WeChat mini-program(IPN-AI group), which supported a closed-loop digital care pathway integrating wound image uploading, AI-assisted staging, stage-matched care recommendations, longitudinal wound tracking, safety alerts, and remote specialist nursing support. Participants were followed for 8 weeks. The primary outcome was caregiver burden measured using the Zarit Burden Interview (ZBI). Secondary outcomes included PI-related service costs, wound improvement rate, and wound area reduction slope. Stratified analyses were performed according to PI stage. <bold>Results</bold> A total of 111 caregiver–patient dyads completed the 8-week follow-up, including 56 in the IPN-AI group and 55 in the IPN group, with a follow-up completion rate of 92.5%. Baseline demographic and wound characteristics were comparable between the two groups ( <italic>P</italic> >0.05). Compared with the IPN group, the IPN-AI group had significantly lower PI-related service costs (495.54 ± 158.80 CNY vs. 854.00 ± 263.70 CNY, <italic>P</italic> <0.001) and a significantly lower total ZBI score at week 8 (48.82 ± 4.94 vs. 68.69 ± 3.80, <italic>P</italic> < 0.001). The overall wound improvement rate was significantly higher in the IPN-AI group than in the IPN group (69.64% vs. 40.00%, <italic>P</italic> =0.002). Stratified analysis showed that the IPN-AI group had significantly higher improvement rates for stage 1 PIs (85.71% vs. 37.50%, <italic>P</italic> =0.009) and stage 2 PIs (74.19% vs. 46.43%, <italic>P</italic> =0.029). However, no statistically significant between-group differences were observed for stage 3 or stage 4 PIs ( <italic>P</italic> >0.05). Analysis of wound area reduction slopes further indicated faster healing in stage 1 and stage 2 PIs in the IPN-AI group, whereas no significant advantage was observed for stage 3–4 PIs. <bold>Conclusion</bold> The IPN-AI pathway further reduced family caregiver burden and PI-related service costs and promoted wound healing in stage 1–2 PIs. The YOLOv8s-assisted PI staging tool appears to be particularly valuable for early PI identification, standardized home care, and continuous monitoring. For complex stage 3–4 PIs, its role may lie mainly in risk alerting, disease tracking, and timely referral support rather than replacing professional wound treatment. These stage-specific effects suggest that the IPN-AI pathway should not be regarded as a homogeneous intervention applicable to all PI stages, but should instead be functionally positioned and stratified according to wound severity. </p>