Evaluation of the Accuracy of Fasting Blood Glucose Measurement Using Wearable Non-invasive Glucose Meters and Their Influencing Factors

doi:10.12114/j.issn.1007-9572.2025.0526

Abstract

Abstract:

Background

Blood glucose monitoring is a core component of glycaemic management. Traditional invasive testing carries risks of pain and infection, while non-invasive blood glucose monitoring technologies have advanced rapidly, with their associated performance metrics awaiting evaluation.

Objective

To investigate the accuracy, wearability, and influencing factors of non-invasive wearable glucose monitors in measuring fasting blood glucose levels, thereby determining their practical applicability.

Methods

Between 13 January and 11 May 2025, a total of 103 participants were recruited from three branches of the Beijing Physical Examination Centre. Fasting blood glucose values were measured simultaneously using wearable non-invasive glucose meters, Roche Accu-Chek Performa, and venous blood sampling, with fingerstick and venous glucose values serving as reference standards. Accuracy was assessed using mean absolute relative error (MARD) and error grid analyses. Device wear comfort was evaluated via questionnaire survey. Least absolute shrinkage and selection operator regression and Logistic regression analyses were used to explore the potential factors influencing the accuracy of wearable non-invasive glucose meters.

Results

The study included 103 participants, 33 males (32.04%) and 70 females (67.96%), aged between 22 and 70 years. When using fingerstick and venous blood glucose as reference standards, the MARD values were 19.79% and 23.99%, respectively. All measurement points fell within the Clarke error grid's A+B zone, with 97.09% and 95.14% of measurement points respectively located in the consensus error grid's A+B zone. Logistic regression analyses showed that accuracy was associated with low-density lipoprotein cholesterol when fingertip glucose was used as the reference (OR=0.80, 95%CI=0.70-0.91, P<0.001), and neutrophil count (OR=0.89, 95%CI=0.81-0.97, P=0.010), platelet volume distribution width (OR=1.05, 95%CI=1.01-1.09, P=0.048). 89.32% (92 subjects) of participants found the wearable non-invasive glucose meter comfortable; those reporting the device uncomfortable had higher mean haemoglobin concentrations (OR=1.04, 95%CI=1.03-1.04, P<0.001).

Conclusion

The wearable non-invasive blood glucose monitor demonstrated moderate measurement accuracy and high wearing comfort in real-world physical examination populations. The accuracy was primarily influenced by low-density lipoprotein cholesterol, neutrophil count, and platelet distribution width, while comfort was associated with MCHC. With further improvements in accuracy, this device holds promise for providing a valuable reference for blood glucose management.

Key words: Fasting blood glucose, Accuracy evaluation, Non-invasive glucose meter, Wearable devices, Comfort

摘要：

背景

血糖监测是血糖管理的核心环节，传统有创检测会引起疼痛和感染风险，无创血糖监测技术迅速发展，其检测性能有待评估。

目的

探讨可穿戴无创血糖仪测量空腹血糖的准确性、佩戴舒适度及影响因素，明确其应用可行性。

方法

2025-01-13—05-11共招募北京市体检中心3个分部的103名受试者，同时分别使用可穿戴无创血糖仪、罗氏卓越金采型血糖仪、静脉采血测量受试者空腹血糖，以指尖和静脉血糖为参考。通过平均绝对相对误差（MARD）、误差栅格分析等评价准确性；通过问卷调查设备佩戴舒适度；采用最小绝对收缩和选择算子回归、Logistic回归分析探讨可穿戴无创血糖仪准确性的影响因素。

结果

共纳入103名受试者，其中男33名（32.04%）、女70名（67.96%），年龄范围为22~70岁。以指尖和静脉血糖为参考时，可穿戴无创血糖仪的MARD分别为19.79%、23.99%，测量点均位于Clarke误差栅格的A+B区，分别有97.09%、95.14%的测量点位于共识误差栅格的A+B区。Logistic回归分析显示以指尖血糖为参考时，准确性与低密度脂蛋白胆固醇（OR=0.80，95%CI=0.70~0.91，P<0.001）有关；以静脉血糖为参考时，准确性与中性粒细胞计数（OR=0.89，95%CI=0.81~0.97，P=0.010）、血小板体积分布宽度（OR=1.05，95%CI=1.01~1.09，P=0.048）有关。89.32%（92例）受试者认为可穿戴无创血糖仪舒适，设备佩戴不舒适者的平均血红蛋白水平较高（OR=1.04，95%CI=1.03~1.04，P<0.001）。

结论

该可穿戴无创血糖仪在真实世界体检人群的应用上表现出一定的测量准确性及较高的佩戴舒适度；准确性主要受低密度脂蛋白胆固醇、中性粒细胞计数及血小板体积分布宽度影响，舒适度与平均血红蛋白浓度有关，继续提升准确性后有望为血糖管理提供参考。

关键词: 空腹血糖, 准确性评价, 无创血糖仪, 可穿戴设备, 舒适度

XIE Wenhan,WANG Jinqi,LIU Hao, et al. Evaluation of the Accuracy of Fasting Blood Glucose Measurement Using Wearable Non-invasive Glucose Meters and Their Influencing Factors[J]. Chinese General Practice, 2026, 29(25): 3624-3630. DOI: 10.12114/j.issn.1007-9572.2025.0526.
谢文涵,王金麒,刘浩等. 可穿戴无创血糖仪测量空腹血糖的准确性及影响因素研究[J]. 中国全科医学, 2026, 29(25): 3624-3630. DOI: 10.12114/j.issn.1007-9572.2025.0526.

Figures/Tables 5

References 27

[1]	Xu Y, Lu J L, Li M, et al. Diabetes in China part 1: epidemiology and risk factors[J]. Lancet Public Health, 2024, 9(12): e1089-e1097. DOI: 10.1016/S2468-2667(24)00250-0.
[2]	Yan Y Q, Wu T T, Zhang M, et al. Prevalence, awareness and control of type 2 diabetes mellitus and risk factors in Chinese elderly population[J]. BMC Public Health, 2022, 22(1): 1382. DOI: 10.1186/s12889-022-13759-9.
[3]	Wang L M, Peng W, Zhao Z P, et al. Prevalence and treatment of diabetes in China, 2013—2018[J]. JAMA, 2021, 326(24): 2498-2506. DOI: 10.1001/jama.2021.22208.
[4]	Seidu S, Kunutsor S K, Ajjan R A, et al. Efficacy and safety of continuous glucose monitoring and intermittently scanned continuous glucose monitoring in patients with type 2 diabetes: a systematic review and meta-analysis of interventional evidence[J]. Diabetes Care, 2024, 47(1): 169-179. DOI: 10.2337/dc23-1520.
[5]	刘宇轩，曹超. 可穿戴柔性连续动态血糖监测技术发展与挑战[J]. 电子技术应用, 2023, 49(12): 75-84. DOI: 10.16157/j.issn.0258-7998.234174.
[6]	程立瑞，朱志刚，张晓峰, 等. 无创血糖监测技术研究进展[J]. 中国医学装备, 2025, 22(5): 160-165. DOI: 10.3969/j.issn.1672-8270.2025.05.031.
[7]	Sunstrum F N, Khan J U, Li N W, et al. Wearable textile sensors for continuous glucose monitoring[J]. Biosens Bioelectron, 2025, 273: 117133. DOI: 10.1016/j.bios.2025.117133.
[8]	Wu J C, Liu Y Q, Yin H, et al. A new generation of sensors for non-invasive blood glucose monitoring[J]. Am J Transl Res, 2023, 15(6): 3825-3837.
[9]	Feng R, Geng Z X, Li A, et al. Noninvasive blood glucose monitor via multi-sensor fusion and its clinical evaluation[J]. Sens Actuat B Chem, 2021, 332: 129445. DOI: 10.1016/j.snb.2021.129445.
[10]	史纯纯，吴燕，靳志涛, 等. 基于人工智能算法的无创血糖检测仪的临床准确性研究[J]. 中国医学装备, 2024, 21(9): 28-32. DOI: 10.3969/j.issn.1672-8270.2024.09.006.
[11]	Zhang Y F, Zhang L L, Wang L, et al. Subcutaneous depth-selective spectral imaging with mμSORS enables noninvasive glucose monitoring[J]. Nat Metab, 2025, 7(2): 421-433. DOI: 10.1038/s42255-025-01217-w.
[12]	He X X, Chang S J, Settu K, et al. An anti-HCT-interference glucose sensor based on a fiber paper-based screen-printed carbon electrode[J]. Sens Actuat B Chem, 2019, 297: 126763. DOI: 10.1016/j.snb.2019.126763.
[13]	Sun T Y, Zhou Y L, Jing Y Z, et al. Micro interstitial fluid extraction and detection device integrated with the optimal extraction conditions for noninvasive glucose monitoring[J]. Biosens Bioelectron, 2023, 237: 115515. DOI: 10.1016/j.bios.2023.115515.
[14]	中华医学会糖尿病学分会，中华医学会感染病学分会，中华医学会组织修复与再生分会. 中国糖尿病足防治指南(2019版)(Ⅴ)[J]. 中华糖尿病杂志, 2019, 11(6): 387-397. DOI: 10.3760/cma.j.issn.1674-5809.2019.06.005.
[15]	Schrangl P, Reiterer F, Heinemann L, et al. Limits to the evaluation of the accuracy of continuous glucose monitoring systems by clinical trials[J]. Biosensors, 2018, 8(2): 50. DOI: 10.3390/bios8020050.
[16]	Battelino T, Danne T, Bergenstal R M, et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range[J]. Diabetes Care, 2019, 42(8): 1593-1603. DOI: 10.2337/dci19-0028.
[17]	Garg N, Lewis K, White P C, et al. Continuous glucose monitor accuracy for diabetes management in hospitalized children[J]. Diabetes Care, 2025, 48(2): 259-264. DOI: 10.2337/dc24-1562.
[18]	孙曾梅，邬云红，张惠勤, 等. 国产无创血糖仪在不同红细胞压积人群检测中的准确性评价[J]. 中国医学装备, 2023, 20(11): 30-35. DOI: 10.3969/J.ISSN.1672-8270.2023.11.007.
[19]	李婷婷，宋雪娇，次仁旺姆, 等. 评估基于代谢热整合法的无创血糖检测仪在高海拔地区性能的研究[J]. 中国糖尿病杂志, 2023, 31(7): 509-513. DOI: 10.3969/j.issn.1006-6187.2023.07.006.
[20]	Zhang B H, Wang L Y, Chang S J, et al. Flexible cellulose paper-based biosensor from inkjet printing for non-invasive glucose monitoring[J]. Polym Test, 2024, 137: 108527. DOI: 10.1016/j.polymertesting.2024.108527.
[21]	Chang T R, Li H, Zhang N R, et al. Highly integrated watch for noninvasive continual glucose monitoring[J]. Microsyst Nanoeng, 2022, 8: 25. DOI: 10.1038/s41378-022-00355-5.
[22]	Zhu W W, Yu H X, Li W J, et al. pH calibration allows accurate glucose detection in interstitial fluid via reverse iontophoresis[J]. Nat Commun, 2025, 16(1): 10413. DOI: 10.1038/s41467-025-65453-0.
[23]	ISO. In vitro diagnostic test systems—Requirements for blood-glucose monitoring systems for self-testing in managing diabetes mellitus: ISO 15197:2013[S]. Geneva: International Organization for Standardization, 2013.
[24]	Joseph J I. Review of the long-term implantable senseonics continuous glucose monitoring system and other continuous glucose monitoring systems[J]. J Diabetes Sci Technol, 2021, 15(1): 167-173. DOI: 10.1177/1932296820911919.
[25]	Wu H M, Fang F, Wang C C, et al. Selective molecular recognition of low density lipoprotein based on β-cyclodextrin coated electrochemical biosensor[J]. Biosensors, 2021, 11(7): 216. DOI: 10.3390/bios11070216.
[26]	Wood K A, Cunegundes P S, De Souza J G, et al. Injury to tissue caused by device penetration of the skin triggers formation of extracellular traps[J]. Commun Biol, 2025, 8: 1669. DOI: 10.1038/s42003-025-09062-z.
[27]	Shokrekhodaei M, Cistola D P, Roberts R C, et al. Non-invasive glucose monitoring using optical sensor and machine learning techniques for diabetes applications[J]. IEEE Access, 2021, 9: 73029-73045. DOI: 10.1109/access.2021.3079182.

基本资料	数据
年龄[M（QR），岁]	26（21）
性别（男/女）	33/70
糖尿病患病[名（%）]	8（7.77）
糖尿病前期[名（%）]	16（15.53）
空腹血糖
静脉血糖[M（QR），mmol/L]	4.86（0.49）
指尖血糖[M（QR），mmol/L]	5.20（0.80）
无创血糖[M（QR），mmol/L]	5.26（1.43）
BMI[M（QR），kg/m²]	22.30（4.40）
腰臀比[M（QR）]	0.81（0.10）
中性粒细胞计数[M（QR），×10⁹/L]	3.27（1.23）
血小板体积分布宽度[M（QR），%]	10.80（2.30）
总胆固醇[M（QR），mmol/L]	4.56（0.94）
甘油三酯[M（QR），mmol/L]	0.94（0.65）
低密度脂蛋白胆固醇[M（QR），mmol/L]	2.63（0.70）
高密度脂蛋白胆固醇[M（QR），mmol/L]	1.45（0.34）
糖化血红蛋白[M（QR），%]	5.40（0.45）

基本资料	数据
年龄[M（QR），岁]	26（21）
性别（男/女）	33/70
糖尿病患病[名（%）]	8（7.77）
糖尿病前期[名（%）]	16（15.53）
空腹血糖
静脉血糖[M（QR），mmol/L]	4.86（0.49）
指尖血糖[M（QR），mmol/L]	5.20（0.80）
无创血糖[M（QR），mmol/L]	5.26（1.43）
BMI[M（QR），kg/m²]	22.30（4.40）
腰臀比[M（QR）]	0.81（0.10）
中性粒细胞计数[M（QR），×10⁹/L]	3.27（1.23）
血小板体积分布宽度[M（QR），%]	10.80（2.30）
总胆固醇[M（QR），mmol/L]	4.56（0.94）
甘油三酯[M（QR），mmol/L]	0.94（0.65）
低密度脂蛋白胆固醇[M（QR），mmol/L]	2.63（0.70）
高密度脂蛋白胆固醇[M（QR），mmol/L]	1.45（0.34）
糖化血红蛋白[M（QR），%]	5.40（0.45）

变量	b	SE	OR（95%CI）	t值	P值
以指尖血糖为参考
低密度脂蛋白胆固醇	-0.226	0.066	0.80（0.70~0.91）	-3.429	<0.001
以静脉血糖为参考
中性粒细胞计数	-0.118	0.045	0.89（0.81~0.97）	-2.635	0.010
血小板体积分布宽度	0.045	0.013	1.05（1.01~1.09）	2.001	0.048
腰臀比	0.814	0.585	2.26（0.71~7.21）	1.391	0.167

变量	b	SE	OR（95%CI）	t值	P值
以指尖血糖为参考
低密度脂蛋白胆固醇	-0.226	0.066	0.80（0.70~0.91）	-3.429	<0.001
以静脉血糖为参考
中性粒细胞计数	-0.118	0.045	0.89（0.81~0.97）	-2.635	0.010
血小板体积分布宽度	0.045	0.013	1.05（1.01~1.09）	2.001	0.048
腰臀比	0.814	0.585	2.26（0.71~7.21）	1.391	0.167