Fig.2 Structural features and catalytic performance of enzyme@COF. (a) Schematic of enzyme encapsulation into the COF matrix. (b) XRD spectrum of the aCOF and COF. (c) FTIR spectrum of the TAPB, DMTA, aCOF, COF, and CA-SOx@COF. (d) N2 adsorption-desorption isotherms and (e) pore size distributions of aCOF, COF, and CA-SOx@COF. (f) UV–vis absorption spectra of CA-SOx@COF-containing reaction solutions with varying Cre concentrations. (g) Relative catalytic activities of free CA-SOx and CA-SOx@COF after exposure to hot water, ethanol, acid solution, and alkaline solution. The activity of untreated free CA-SOx was referenced as 100%, and error bars are standard deviations (n = 3). (h) Storage stability of CA-SOx@COF compared with free CA-SOx in a homogeneous solution under identical conditions. UV–vis absorption spectra of the reaction solutions containing (i) SOx@COF at different Sar concentrations and (j) LOx@COF at different LA concentrations.#

在传感电极阵列设计方面，如图3所示，团队采用丝网印刷技术构建了五电极系统，引入氨基化MXene/单壁碳纳米管与普鲁士蓝介体以优化电子传递。针对汗液中内源性肌氨酸引起的信号重叠，系统设计了专用肌氨酸电极实现实时背景校准。校准后的肌酸传感器在1至200 μM范围内呈现优异线性，灵敏度达到24.5 nA μM⁻¹，检出限低至0.54 μM；乳酸传感器在1至50 mM生理浓度范围内同样表现出高选择性与重复性。

Fig.3 Design, electrochemical characterization, and analytical performance of the sensing modules. (a) Scheme of the Cre, Sar, and LA sensors. The (b) CV and (c) EIS responses of the carbon electrode, MXene/SWCNTs electrode, PB-modified electrode, and Cre sensor in 5 mM K3[Fe(CN)6] containing 0.1 M KCl solution. Scan rate: 50 mV s−1. (d) Schematic illustration of the detection principle of the Cre sensor based on CA-SOx@COF. (e) Amperometric responses of the Cre sensors for different concentrations of Cre detection. Inset： corresponding calibration plot with a linear fit. (f) Selectivity of the Cre sensor was evaluated by continuously adding Glu, AA, UA, LA, NaCl, KCl, and Cre. (g) Reproducibility of six Cre sensors measured in 1 µM Cre solution. (h) Amperometric responses of the Sar sensors for different concentrations of Sar detection. Inset, corresponding calibration plot with a linear fit. (i) Selectivity of the Sar sensor was evaluated by continuously adding Glu, AA, UA, Cre, LA, NaCl, and Sar. (j) Response of the Cre sensor under varying Sar levels. (k) Amperometric responses of the Cre sensor in the presence of Sar before and after Sar correction. (l) Schematic illustration of the detection principle of the LA sensor based on LOx@COF. (m) Amperometric responses of the LA sensors for different concentrations. Inset, corresponding calibration plot with a linear fit. (n) Selectivity of the LA sensor was evaluated by continuously adding Glu, AA, UA, Cre, Sar, NaCl, and LA. (o) Reproducibility of six LA sensors measured in a 5 mM LA solution.#

在汗液流体调控方面，如图4所示，微流控芯片结合了锯齿形侧壁与仿特斯拉阀不对称结构。多物理场仿真与染色实验证实，该结构产生了方向依赖的水力阻力，在低至0.2 μL min⁻¹的极低汗液流速下依然能实现高效单向输运，最大流动阻力仅约10 Pa，有效抑制了液体回流与前后样本交叉污染。

Fig.4 Structural design and flow analysis of the microfluidic chips. (a) Schematic of the multi-inlet microfluidic chip. Optical images of (b) the inlet and (c) the designed microchannel. (d) Theoretical simulation results of (i) velocity and (ii) pressure distribution within the microchannel. (e) Theoretical simulation results of sweat flow in microchannels. (f) Schematic of the patterned microchannel with tunable geometrical parameters. Theoretical simulation results of (g) sweat flow and (h) flow velocity in microchannels with different height ratios. Theoretical simulation results of (i) sweat flow and (j) flow velocity in microchannels with different width ratios. (k) Schematic of the unidirectional flow principle. (l) Theoretical calculations comparing the flow rates of conventional microchannels and designed microchannels. (m) Experimental demonstration of fluid flow in the microfluidic chip driven by a syringe pump at 3 µL min−1.#

在系统级集成与人体试验验证中，如图5所示，微型化电路模块通过蓝牙低功耗技术实现了信号实时解调与手机APP端浓度转换，在健康志愿者前臂原位汗液诱导测试中实现了稳定读数。

Fig.5 Schematic diagram and performance of an integrated wearable sweat platform for sweat monitoring. (a) The block diagram of the electronic system of the sensing platform. RF, radio frequency; CPU, central processing unit; ADC, analog-to-digital converter; DAC, digital-to-analog converter; SPI, serial peripheral interface; SoC, system on chip. (b) Optical image of (i) a flexible sensor array fabricated via screen printing; (ii) an integrated microfluidic chip with a sensing platform; (iii) a microcircuit board; and (iv) a sensor-wearing diagram. Scale bars: 10 mm for panels (i–iv). (c) Customized mobile application for real time Cre, Sar, and LA levels tracking. (d) On-body real time perspiration analysis.#

如图6所示，人体生理扰动实验表明，汗液与血清中的肌酸及乳酸浓度变化趋势高度吻合，皮尔逊相关系数分别达到0.751和0.904，且传感器读数与质谱金标准高度一致。在初步临床样本验证中，慢性肝病患者组的汗液肌酸水平显著低于健康对照组，而乳酸水平显著升高，充分验证了该无创监测方案区分代谢异常的可行性。

Fig.6 Human-subject validation of the wearable sensor for noninvasive sweat monitoring of Cre and LA. (a) Metabolic pathway of Cre and LA, along with serum and sweat fluctuations caused by exogenous intake. (b–d) Temporal changes in sweat and serum Cre and LA levels from subject #1 before and after exogenous Cre intake. (e,f) Temporal changes in sweat and serum Cre and LA levels from subject #2 before and after running. Correlations between sweat and serum (g) Cre and (h) LA levels. Lines represent the fitted trendlines. Sensor-measured (i) Cre and (j) LA concentrations in sweat samples versus corresponding LC-MS readouts. Data were measured from all sweat samples collected in the above evaluation experiment. The line represents the linear-fitted trendline. (k–n) Sweat Cre and LA levels in healthy subjects and patients with chronic liver disease (n = 10 per group). (k, m) Sweat Cre and LA concentrations; error bars represent the standard deviation. (l,n) Corresponding box-and-whisker plots showing significantly higher Cre levels in healthy subjects (two-tailed Wilcoxon rank-sum test, W = 152, ***p < 0.001; exact p = 7.58 × 10−5) and higher LA levels in patients with chronic liver disease (W = 145.5, **p < 0.01; exact p = 0.0011). Boxes indicate the interquartile range, center lines indicate the median, and whiskers indicate the minimum and maximum non-outlier values. Individual measurements are shown as scattered points.#