Telemedicine in ER Care via 5G AR: Time Series Analysis

Yao Zhang, Bangbo Xia, Bin Zheng, Yucai Hong

Keywords: Augmented Reality, Tele-mentoring, Emergency Care, Time Series Analysis

Abstract:

INTRODUCTION
Compared with metropolitan centres, rural communities have fewer healthcare resources, creating substantial barriers to timely care, especially in emergency settings, where minutes matter and preventing avoidable delays can markedly improve outcomes. Our earlier analyses suggested that deploying the 5G-based Remote Healthcare System (5G-RHS) improved several process measures; however, cross-sectional pre/post contrasts can be biased by secular trends and seasonality. To isolate changes attributable to 5G-RHS beyond background variation, we analysed the full continuous time series using an interrupted time series (ITS) design centered on the actual rollout. One hospital contributed a complete pre- and post-intervention series, while other sites provided post-only context. Our objective was to estimate both the immediate level change at implementation and the post-rollout slope change in key ER metrics, and to assess whether 5G-RHS helps rural clinicians deliver care closer to metropolitan standards.

MATERIALS AND METHODS
We analysed multiple clinical measures in trauma cases, including Early Warning rate, FAST (Focused Assessment with Sonography for Trauma) time, transfusion preparation time, total case volume, and mortality rate. We fit an ITS of the form:
Outcome ~ b0 + b1×Post + b2×TimeIndex + b3×(Post×TimeIndex) + Month fixed effects.
We reported coefficients with standard errors, focusing on the level change b1 (immediate shift at rollout) and the slope change b3 (difference in monthly trend). We also estimated a pooled fixed-effects model with hospital and month indicators to provide cross-site context.

RESULTS AND DISCUSSION
ITS revealed a significant immediate increase in the Early Warning rate with no subsequent slope change, a step-up that was sustained. FAST time showed a significant immediate decrease without additional post-rollout trend change. Total and severe-case measures exhibited no immediate level shift but a significant downward post-rollout slope, consistent with progressive month-over-month improvement under 5G-RHS. ICU length of stay and severe-case mortality remained same fluctuation pattern. Table 1 summarizes the ITS estimates and standard errors; Figure 1 visualizes the observed series alongside the fitted pre/post segments.

CONCLUSIONS
ITS indicates that AR-assisted 5G-RHS produced a sharp change in early recognition of patient condition (an immediate level increase in Early Warning percent) and a sustained improvement in time-critical workflows (a month-over-month decrease in transfusion preparation time). For some measures like ICU length of stay and severe-case mortality, implementation of 5G-RHS did not show significant impacts. These gains support AR-enabled tele-mentoring as a scalable approach to narrow rural–urban quality gaps in emergency care.

Submission Number: 52