The advent of digital health technologies has transformed chronic disease management, with diabetes care positioned at the forefront of this revolution [1,2]. Mobile health apps, wearable devices, and digital biomarkers now offer unprecedented opportunities for real-time patient monitoring and personalized intervention strategies [3-7]. However, despite significant advances in glucose monitoring technologies, exercise compliance assessment—a cornerstone of diabetes management—remains predominantly dependent on subjective self-reporting methods that lack the precision and objectivity demanded by contemporary precision medicine approaches [8,9]. Current digital health ecosystems for diabetes management excel in glucose tracking but demonstrate a critical deficiency in objective exercise assessment [1,10,11]. While continuous glucose monitors provide minute-by-minute glycemic data, exercise compliance monitoring continues to rely on patient diaries, smartphone step counters, or intermittent subjective assessments during clinical encounters. This disparity creates a fundamental imbalance in the digital health toolkit available to primary care providers, who must make evidence-based treatment decisions regarding exercise therapy optimization without access to reliable, objective compliance data [12].

Type 2 diabetes affects over 460 million individuals worldwide, with substantial morbidity, mortality, and health care costs [13-16]. Despite therapeutic advances, only 50% to 60% of patients achieve hemoglobin A_1c (HbA_1c) targets in real-world practice, creating urgent challenges for primary care diabetes management [17,18].

In primary care settings, exercise therapy and dietary modification are universally recommended as first-line interventions [19-24]; however, the digital infrastructure supporting these recommendations remains incomplete. While dietary tracking applications and digital nutrition platforms have achieved widespread adoption, exercise monitoring capabilities lag significantly behind, creating an asymmetric digital health ecosystem that inadequately supports comprehensive lifestyle intervention strategies [3,9].

Traditional methods of monitoring exercise compliance, including patient self-reports, activity diaries, and subjective assessments, are notoriously unreliable and provide insufficient objective data for clinical decision-making [12]. This monitoring deficit creates a critical gap in primary care: physicians cannot provide appropriately tailored exercise therapy guidance without reliable evidence of patient exercise implementation [25].

Self-care management could potentially be enhanced through objective body composition assessment as evidence of exercise implementation. Bioelectrical impedance analysis (BIA) offers promise for detecting exercise-induced physiological changes, providing objective markers that could inform clinical decision-making [26,27]. However, fundamental uncertainties regarding optimal measurement parameters severely limit clinical use [28].

Body composition assessment reveals diverse reactance frequency characteristics across the spectrum from 5 kHz to 250 kHz, with each frequency potentially providing different physiological information [26,29]. Similarly, anatomical measurement sites—from individual limbs to whole-body assessment—may exhibit differential sensitivity to exercise-induced changes [30,31]. The critical clinical question remains: which specific frequency and anatomical site combination most reliably reflects exercise implementation and predicts treatment success in real-time tracking applications?

Recent systematic evaluations reveal the extent of contradictions in bioelectrical impedance research [28,32]. Studies have shown conflicting results regarding which body segments provide optimal metabolic information, with some studies favoring upper extremity measurements and others showing no regional differences [30,33]. The comprehensive literature examining body composition methodologies demonstrates the need for standardized approaches [34].

We propose that these contradictions reflect a previously unrecognized phenomenon: culturally imposed localized inactivity. Across global populations, food preparation and consumption practices create systematic asymmetric limb use patterns that represent “manufactured” localized disuse. For instance, cultures using utensils consistently favor one hand for manipulation while the other provides stabilization, creating daily repetitive asymmetric loading patterns. This cultural specificity of body usage represents a unique form of environmental laterality conditioning that has been overlooked in previous bioimpedance research.

We hypothesize that anatomical specificity of metabolic biomarkers depends on the degree of baseline body usage asymmetry in the studied population. This unified framework predicts the following:

This framework enables systematic reconciliation of contradictory findings while providing practical guidance for optimizing digital biomarker protocols in diabetes care.

Given this theoretical framework and the documented need for objective physical activity monitoring in primary care diabetes management, this study aimed to: (1) validate a digital bioimpedance protocol for objective physical activity detection in Japanese adults with type 2 diabetes, (2) identify population-specific anatomical sites demonstrating optimal sensitivity for activity tracking, and (3) evaluate the clinical use of this framework for predicting glycemic target achievement in real-world primary care settings. We hypothesized that left-arm 50-kHz reactance would demonstrate superior predictive value compared to other anatomical sites, consistent with high-asymmetry population patterns.

This quasi-experimental study follows the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines (Checklist 1) for reporting observational studies. The study used temporal allocation (sequential enrollment by time period) rather than individual randomization, which is acknowledged as a significant limitation affecting causal inference. For more details, see the “Study Limitations” section.

This quasi-experimental, temporally allocated observational study was conducted at Saiseikai Maebashi Hospital, Japan, from January 2022 to July 2023. The study was approved by the institutional review board of Gunma Paz University (IRB No PAZ23-9; approval date: June 15, 2021) and was retrospectively registered in the UMIN Clinical Trials Registry (UMIN000058452; registration date: July 14, 2025). The rationale for retrospective registration is disclosed transparently given the observational design. This study followed the STROBE guidelines for reporting observational studies.

Adults aged 20 to 80 years with type 2 diabetes (HbA_1c 7.0%‐10.0%) were recruited through the endocrinology clinic. The study used an opt-out consent process approved by the institutional review board; patients could decline participation at any time. Japanese adults were selected as an ideal population for testing the unified framework, given consistent asymmetric upper-limb use in daily activities.

The comprehensive intervention (period A) included individualized exercise prescriptions based on baseline fitness assessment. Recommended activities included brisk walking (primary modality), cycling, swimming, or home-based calisthenics, prescribed at moderate intensity (50%‐70% of age-predicted maximum heart rate or 12‐14 on the Borg Rating of Perceived Exertion scale). Target duration was 30 minutes per session, 5 days per week, with gradual progression from baseline activity levels. Participants received written exercise logs and were instructed to record activity type, duration, and perceived exertion. The partial intervention (period B) received standard verbal advice about exercise benefits without individualized prescriptions, activity logs, or follow-up counseling. Standard care (period C) received routine diabetes education consistent with usual clinic practice, which included general encouragement for physical activity but no structured exercise prescription or monitoring.

Hand dominance and use asymmetry were assessed using the Edinburgh Handedness Inventory administered at baseline. Participants reported: (1) writing hand preference, (2) hand used for complex manipulations during eating with utensils (chopsticks, spoon, and fork), and (3) age when these patterns became established. Data were collected via standardized questionnaire.

As registered in UMIN000058452, participants were allocated to treatment groups based on distinct enrollment periods, representing a quasi-experimental design with temporal allocation. This nonrandomized design was chosen for pragmatic implementation feasibility in routine clinical practice as follows:

The primary predictive analysis employed multivariable logistic regression with HbA_1c <7% achievement as the outcome. Candidate predictors included baseline bioimpedance measurements (segmental resistance and reactance at each frequency), demographic variables (age, sex, and BMI), diabetes characteristics (duration, baseline HbA_1c, and medication regimen), and environmental factors (neighborhood walkability score). We examined interactions between left-arm 50-kHz reactance and walkability score based on our a priori hypothesis of environment-individual synergy. Model discrimination was assessed using the area under the receiver operating characteristic curve (AUC), and optimal classification threshold was determined using the Youden index. Internal validation used 10-fold cross-validation to assess overfitting; the cross-validated AUC differed from the apparent AUC by <0.02, indicating minimal overfitting. Feature selection was guided by clinical plausibility and our theoretical framework rather than purely data-driven algorithms, minimizing risk of spurious associations from multiple testing.

As a sensitivity analysis, we applied inverse probability weighting (IPW) with stabilized weights estimated from a logistic regression including age, BMI, baseline HbA_1c, calendar period (as dummy variables), medication indicators (metformin, DPP-4i, SGLT2i, GLP-1 RA, and insulin), and Walk Score. HC3 robust standard errors were used to account for heteroskedasticity. IPW was used to assess whether the association between left-arm reactance change and HbA_1c outcomes persisted after rigorous adjustment for measured confounders.

During the study period, 218 individuals were screened and 25 were excluded. A total of 196 participants underwent temporal allocation to one of 3 care periods: period A (comprehensive intervention, n=65), period B (partial intervention, n=31), and period C (standard care, n=100). Follow-up completion was high; one participant in period B relocated outside the catchment area and one in period C withdrew consent (Figure 1). Primary analyses used an intention-to-treat approach including all allocated participants; a per-protocol sensitivity analysis excluded participants with incomplete outcome data (period B: n=29; period C: n=87). Baseline characteristics were similar across all 3 periods (Table 1).

Notes. Continuous variables are mean ± SD; categorical variables are n (%). P values from one-way ANOVA (continuous), Kruskal–Wallis (non-normal), and χ² tests (categorical) comparing three periods. P values are unadjusted and provided for descriptive comparability; standardized mean differences are presented in Supplementary Table S1 in Multimedia Appendix 1. This study used a temporally allocated, nonrandomized design; therefore, baseline tests do not imply randomized balance. Walk Score was obtained via the public API in January–February 2022 for participants' residential addresses. In multivariable models, the association of left-arm 50 kHz reactance with HbA_1c goal attainment remained significant after FDR correction (adjusted P=.044).

Across 4 months, HbA_1c declined in all groups, with the largest improvement in period A, intermediate in period B, and minimal change in period C (Figure 2). The group-wise separation was apparent by 2 months and persisted to 4 months, consistent with a dose-response pattern aligned with intervention intensity.

Segmental reactance at 50 kHz showed anatomy-specific and outcome-dependent trajectories (Figure 3). In the right arm, reactance decreased in participants who ultimately achieved the HbA_1c target but increased among those who did not. In contrast, in the left arm, reactance increased among target achievers and decreased among nonachievers. These opposing directions indicate a significant interaction between limb side and glycemic target status, with the left arm demonstrating a pattern consistent with the a priori primary signal.

When expressed as net change (time 3 to time 1), left-arm reactance increased among achievers and declined among nonachievers, whereas right-arm reactance showed the reverse (Figure 4). The directionality was concordant with the longitudinal trajectories above and robust across intervention groups.

In exploratory analyses combining neighborhood walkability with segmental bioimpedance, patterns were consistent with a walkability-dependent association; full estimates are reported in the “Interaction: Walk Score×Left-Arm 50-kHz Reactance“ subsection and Figures 3 and 4.

In the multivariable model including an interaction between left-arm reactance at 50 kHz (per SD) and neighborhood walkability (per 10-point Walk Score), the interaction term was statistically significant (P=.028). Only the primary estimates are presented in the abstract. Model-based marginal effects indicated that the association between left-arm reactance and HbA1c goal attainment was weaker at lower walkability and stronger at higher walkability. Numerical coefficients and confidence intervals are omitted by request; the visualization is provided in Figures 3–4.

The digital bioimpedance tracking framework demonstrated acceptable performance suitable for real-world clinical implementation, confirming the theoretical predictions. Only left-arm 50-kHz reactance demonstrated predictive value for diabetes treatment success, while all other anatomical sites—including the habitually dominant right arm—showed no association with outcomes.

Among participants with available baseline and follow-up bioimpedance measurements (n=65), the digital framework's ability to provide immediate risk stratification represents a practical step toward objective monitoring over traditional monitoring approaches. The algorithm achieved an AUC of 0.847 (95% CI, 0.789-0.905), indicating excellent discrimination ability for identifying patients likely to achieve glycemic targets within the critical early intervention window. Using the Youden-optimal threshold, overall accuracy was 84%; corresponding sensitivity and specificity are reported in Multimedia Appendix 1.

HbA_1c <7% achievement demonstrated clear intervention intensity gradient reflecting the temporal allocation design:

The framework’s prediction accuracy was maintained across different measurement time points, with 2-month predictions showing 84% concordance with 4-month outcomes. This temporal stability is essential for clinical implementation, as it enables health care providers to make evidence-based treatment adjustments during the therapeutic window when intervention modifications are most effective.

This quasi-experimental validation study demonstrates that digital bioimpedance analysis can objectively detect physical activity in adults with type 2 diabetes. Among 196 participants across 3 intervention intensities, only left-arm 50-kHz reactance predicted glycemic target achievement (adjusted P=.044), while all other anatomical sites showed no association. The predictive algorithm achieved excellent discrimination (AUC=0.847; 84% accuracy), with clear dose-response relationship across intervention intensities (χ²₄ for trend=14.23; P<.001). These findings validate population-specific biomarker selection and provide a practical framework for objective activity monitoring in primary care settings.

BIA exploits the frequency-dependent electrical properties of biological tissues to noninvasively assess body composition and cellular function. When alternating current is applied to tissue, the measured impedance comprises 2 components: resistance (R), reflecting the opposition to current flow through ionic conductors (intra- and extracellular fluids), and reactance (Xc), reflecting the capacitive properties of cell membranes and tissue interfaces that temporarily store electrical charge. The frequency dependence of these properties arises from the differential permeability of cell membranes to current at different frequencies.

At low frequencies (<10 kHz), alternating current cannot penetrate intact cell membranes and flows primarily through extracellular pathways, making measurements predominantly sensitive to extracellular fluid volume. At high frequencies (>100 kHz), current penetrates cell membranes, allowing assessment of total body water (intracellular plus extracellular). Critically, the intermediate frequency of 50 kHz provides a theoretical “sweet spot” where current partially penetrates cell membranes, rendering reactance measurements particularly sensitive to cell membrane integrity, intracellular fluid distribution, and the interface between intracellular and extracellular compartments. This frequency has been extensively validated for assessing cellular health, with decreased reactance indicating compromised cell membrane function or altered fluid distribution characteristic of various pathological states.

The link between physical activity and 50-kHz reactance changes reflects well-established physiological adaptations to habitual exercise. Regular physical activity induces multiple coordinated changes in skeletal muscle tissue that collectively alter bioimpedance properties. First, exercise training enhances muscle cell membrane integrity and ion channel function, increasing the capacitive properties measurable as increased reactance. Second, training promotes intracellular fluid accumulation through increased muscle glycogen storage (each gram of glycogen binds approximately 3 g of water) and enhanced protein synthesis, expanding the intracellular compartment. Third, exercise-induced muscle hypertrophy increases total muscle mass while maintaining or improving cellular organization, thereby increasing tissue reactance per unit volume [46-50].

Importantly, these adaptations develop gradually over weeks to months of habitual activity and persist during periods of measurement (unlike acute exercise–induced changes that dissipate within hours). This temporal stability enables detection during static bioimpedance assessments performed days or weeks after individual exercise bouts, analogous to how chronic training-induced changes in skeletal muscle cross-sectional area or capillary density remain measurable long after acute exercise sessions. Previous investigations have demonstrated that 8 to 12 weeks of regular physical activity significantly increases segmental reactance, with effect sizes comparable to those observed in our study [46,47,50,51].

The exclusive predictive value of left-arm (nondominant) 50-kHz reactance, while right-arm measurements showed no association despite identical measurement protocols, requires mechanistic explanation grounded in principles of muscle physiology and adaptation. We propose a “differential adaptation capacity” hypothesis specific to high-asymmetry populations (such as Japanese adults with consistent right-hand dominance for fine motor tasks and eating utensils).

In populations with pronounced limb-use asymmetry, the dominant limb (right arm) operates near its adaptive ceiling due to high baseline daily activity (writing, eating, occupational tasks). The nondominant limb (left arm), conversely, maintains substantial adaptive reserve capacity. When individuals initiate or increase whole-body physical activity (walking, cycling, and housework), both limbs experience increased activity, but the relative increase is proportionally larger for the previously underused nondominant limb. This differential proportional stimulus may trigger more robust adaptive responses in the nondominant limb, manifesting as detectable reactance increases.

Supporting this hypothesis, biomechanical studies demonstrate that even “symmetrical” activities like walking involve subtle but consistent asymmetries in arm swing and postural stabilization, with the nondominant arm often contributing more to balance and load-bearing. In addition, neurophysiological evidence indicates that motor learning and neuromuscular adaptation proceed more rapidly in less-trained muscle groups, suggesting the nondominant limb may exhibit heightened sensitivity to training stimuli. Critically, this mechanism is population-specific. In populations with lower baseline asymmetry (more balanced bilateral limb use), dominant and nondominant limbs would not exhibit this differential sensitivity, potentially explaining contradictory findings in the bioimpedance literature across different populations and potentially invalidating left-arm measurement protocols in populations with different use patterns.

The observed association between left-arm reactance increases and HbA_1c target achievement likely reflects the known mechanisms linking physical activity to improved glycemic control. Skeletal muscle accounts for approximately 80% of insulin-stimulated glucose uptake, and exercise training enhances multiple aspects of muscle glucose metabolism: increased GLUT4 transporter expression and translocation, improved insulin signaling cascade function, enhanced mitochondrial density and oxidative capacity, and increased muscle glycogen storage capacity. Each of these adaptations contributes both to improved glycemic control and to altered muscle bioimpedance properties (through changes in membrane protein composition, intracellular fluid distribution, and cellular architecture). The anatomical specificity (left arm only) suggests that segmental bioimpedance changes serve as surrogate markers for systemic metabolic improvements rather than directly causing glycemic changes. That is, left-arm reactance changes indicate that meaningful physical activity has occurred (sufficient to induce detectable muscle adaptations), and this same physical activity drives the well-established metabolic benefits that improve glycemic control. The left-arm measurement effectively functions as an objective, passive sensor of habitual activity patterns, superior to self-report but capturing the same underlying behavioral construct [52].

From a health informatics perspective, this study demonstrates the potential of leveraging digital bioimpedance data to bridge the gap between objective physiological measurements and actionable clinical insights. The integration of multi-frequency bioimpedance analysis with environmental data (Walk Score) and real-time clinical decision support represents a practical application of health informatics principles—transforming raw digital signals into clinically meaningful information that can guide personalized diabetes management strategies. The observed environment-person interaction, detected through automated bioimpedance analysis, exemplifies how digital health technologies can uncover complex relationships between individual physiology, behavioral patterns, and environmental contexts that would be difficult to identify through traditional clinical assessment alone. As health informatics continues to evolve toward precision medicine and personalized care, this framework offers a scalable model for integrating readily available digital biomarkers into routine clinical workflows, potentially enabling more responsive and individualized chronic disease management across diverse healthcare settings.

This study represents a practical step from subjective to objective, digitally supported assessment in primary care diabetes management. The validation of population-specific bioimpedance protocols addresses a critical gap in the digital health ecosystem, providing health care providers with evidence-based tools for exercise therapy optimization that complement existing glucose monitoring technologies [5-7,10,11].

The predictive analysis of left arm reactance change provides compelling evidence for its role as a digital biomarker of behavioral adherence. The observation that left arm reactance negatively correlates with HbA_1c change specifically in target achievers (r=−0.392; P=.032), but not in nonachievers (r=0.221; P=.074), suggests this biomarker captures genuine behavioral engagement rather than mere passive physiological drift. The right arm showed the opposite pattern—nonsignificant in achievers but significant in nonachievers—supporting our hypothesis that dominant arm usage patterns differ systematically between those who successfully modify behavior and those who do not. While the modest odds ratio (1.025 per 1 Ω increase) might suggest limited clinical use when considered alongside the dominant baseline HbA_1c predictor, this reflects the complex multivariate nature of glycemic control rather than absence of biological significance. The arm-specific divergence in correlation patterns across achievement strata cannot be explained by confounding or measurement noise and instead points to the nuanced ways bioimpedance captures habitual movement patterns. From a health informatics perspective, this demonstrates how readily available digital signals—collected automatically during routine assessments—can reveal behavioral phenotypes that would otherwise remain invisible in traditional clinical workflows.

The framework’s reliance on standard, FDA 510(k)-cleared equipment eliminates typical implementation barriers associated with novel health technologies. Unlike specialized devices or custom applications that require extensive training and infrastructure investment, the InBody 770 analyzer is already available in many health care settings, enabling immediate adoption without significant capital expenditure or workflow disruption.

This unified framework successfully reconciles previously contradictory bioelectrical impedance findings by demonstrating that anatomical specificity depends on population-specific body usage patterns rather than fixed anatomical relationships. Studies showing right-arm predominance, bilateral symmetry, or no anatomical specificity can now be understood as different points on a continuum of body usage asymmetry.

This framework transforms apparently conflicting literature into complementary evidence supporting the broader principle that baseline activity patterns determine optimal digital biomarker localization for exercise tracking applications.

The discovery that left-arm 50 kHz reactance serves as a population-specific digital biomarker for exercise compliance represents a novel application of precision medicine principles to lifestyle intervention monitoring. This finding challenges the traditional “one-size-fits-all” approach to biomarker development and provides evidence for population-tailored digital health strategies.

The framework’s ability to integrate individual physiological responses with environmental data (neighborhood walkability) demonstrates the potential for multidimensional digital health approaches that consider both personal and contextual factors in treatment optimization. This holistic approach aligns with emerging concepts in digital epidemiology and social determinants of health research [37,38].

The digital framework’s 84% prediction accuracy enables health care providers to make informed decisions about exercise therapy optimization versus medication intensification within weeks rather than months, significantly improving treatment efficiency through objective monitoring. This capability represents a practical step toward objective monitoring in primary care diabetes management, where treatment decisions must be made efficiently within time-constrained consultation environments.

The population-specific approach validated in this study provides a foundation for broader digital health implementation across diverse populations. By demonstrating that optimal bioimpedance protocols depend on cultural body usage patterns rather than fixed anatomical relationships, this research establishes a framework for developing population-tailored digital health solutions that can be adapted for different ethnic groups and geographic regions.

Future integration with smartphone applications, electronic health records, and telemedicine platforms could extend the framework’s use beyond traditional clinical encounters. Real-time bioimpedance data could be transmitted to health care providers, enabling proactive intervention adjustments and reducing the need for frequent in-person visits.

These potential period effects substantially limit our ability to make strong causal inferences about the intervention’s effectiveness. Supplementary sensitivity analyses using ANCOVA and propensity score matching revealed no significant linear trend across intervention intensity levels (P=.25), suggesting that the observed gradient pattern may reflect temporal confounding rather than true dose-response effects. While propensity score matching identified a significant difference between comprehensive intervention and standard care (matched difference: −0.73%; P=.011), this finding cannot definitively distinguish intervention effects from period effects due to the perfect confounding of treatment assignment with time.

The finding that partial intervention participants in high-walkability neighborhoods achieved superior glycemic control (HbA_1c 6.50% vs 9.20% in low-walkability areas) suggests that moderate intervention intensity combined with supportive built environments creates optimal conditions for sustained physical activity. This interaction effect—mechanistically plausible and not explained by seasonal or temporal confounding—supports the validity of our proposed framework for detecting habitual activity through bioimpedance changes.

This environment-person interaction strengthens the biological plausibility of our left-arm reactance findings in three ways. First, if reactance changes truly reflect habitual activity adaptations rather than measurement artifacts or nonspecific effects, we would expect environmental facilitators (walkability) to moderate this relationship—precisely what we observed. Second, the interaction specifically manifested in the partial intervention group, suggesting a “sweet spot” where moderate behavioral support combined with environmental opportunity optimizes activity patterns. Third, the absence of interaction in standard care (where minimal behavioral support was provided) supports the hypothesis that bioimpedance-detected changes require both motivation (intervention) and opportunity (environment).

From a public health perspective, this finding has important implications beyond the methodological limitations of temporal allocation. Even if the gradient across intervention intensities partly reflects period effects, the environment-person interaction demonstrates that neighborhood walkability is a modifiable factor that can be leveraged to optimize diabetes management outcomes. Clinicians can use Walk Score assessment (freely available at walkscore.com) to identify patients whose residential environments may facilitate or hinder physical activity recommendations, enabling more personalized intervention strategies[1,36,41].

The mechanistic evidence from the Walk Score interaction thus serves two purposes: (1) it provides validation for the bioimpedance framework that transcends the study’s temporal allocation limitations and (2) it identifies actionable environmental factors that moderate intervention effectiveness in real-world settings. While we acknowledge that definitive causal inference regarding intervention intensity requires randomized designs, the observed environment-person synergy represents genuine scientific contribution independent of the temporal allocation issue.

Future work includes the following:

This study establishes a validated digital bioimpedance tracking framework that transforms exercise compliance assessment from subjective reporting to objective, real-time measurement in primary care diabetes management. The population-specific approach—demonstrating that left-arm 50 kHz reactance provides optimal sensitivity in high-asymmetry populations— represents a practical step toward objective monitoring in precision digital health, offering evidence-based protocols for immediate clinical implementation using standard equipment.

The framework’s integration of individual physiological responses with environmental data achieves 84% prediction accuracy, providing health care providers with acceptable performance in identifying patients likely to achieve glycemic targets. This capability enables evidence-based treatment optimization within the critical early intervention window, improving the efficiency and effectiveness of primary care diabetes management.

Most importantly, this research validates the concept of population-tailored digital biomarkers, challenging traditional approaches to health technology development and providing a foundation for culturally-adapted digital health solutions. The immediate implementability using FDA 510(k)-cleared equipment eliminates typical barriers to digital health adoption, enabling rapid scaling across health care systems.

Future integration with electronic health records, telemedicine platforms, and mobile apps could extend this framework’s utility, supporting the evolution toward comprehensive, digitally-supported diabetes care that combines real-time monitoring with personalized intervention strategies. This represents a crucial step toward realizing the full potential of digital health technologies in chronic disease management.