A Metamaterial Computational Multi‐Sensor of Grip‐Strength Properties with Point‐of‐Care Human‐Computer Interaction

Abstract Grip strength is a biomarker of frailty and an evaluation indicator of brain health, cardiovascular morbidity, and psychological health. Yet, the development of a reliable, interactive, and point‐of‐care device for comprehensive multi‐sensing of hand grip status is challenging. Here, a relation between soft buckling metamaterial deformations and built piezoelectric voltage signals is uncovered to achieve multiple sensing of maximal grip force, grip speed, grip impulse, and endurance indicators. A metamaterial computational sensor design is established by hyperelastic model that governs the mechanical characterization, machine learning models for computational sensing, and graphical user interface to provide visual cues. A exemplify grip measurement for left and right hands of seven elderly campus workers is conducted. By taking indicators of grip status as input parameters, human‐computer interactive games are incorporated into the computational sensor to improve the user compliance with measurement protocols. Two elderly female schizophrenic patients are participated in the real‐time interactive point‐of‐care grip assessment and training for potentially sarcopenia screening. The attractive features of this advanced intelligent metamaterial computational sensing system are crucial to establish a point‐of‐care biomechanical platform and advancing the human‐computer interactive healthcare, ultimately contributing to a global health ecosystem.


Supplementary Note 1: Mathematical description of voids for MM-F and MM-C
The representative volume elements (RVEs) for the two metamaterials (MM-F and MM-C) considered in this work are shown in Figure S1a.The outline of the voids for the two metamaterials can be expressed by Fourier series expansion:  =   cos  ,  =   sin  with   =  0 [1 +  1 (4) +  2 (8)] where  ranges from 0 to 2π, and three parameters,  1 ， 2 and  0 control the shape of the pore (61,62).For MM-C, as shown by the dotted line, there are For MM-F, there are  1 = 0.11， 2 = −0.05 The side length of the metamaterial is designed to be 80mm to fit the size of a human palm and 4×4 cells are sufficient to reflect the mechanical behavior of the metamaterial and facilitate the integration of the piezoelectric films.Therefore, the square length  0 take 20 mm.The porosity ф 0 for this work is 0.5.And  0 can be determined by the porosity ф 0 :        Supplementary Note 6: The dimensionality reduction and reconstruction algorithms.
Principal component analysis (PCA), aiming at obtaining a hyperplane to express sample points in orthogonal attribute space, is the most commonly used method for dimensionality reduction.
We first centralize all samples X, then calculate the covariance matrix of the samples and perform the eigenvalue decomposition.Then the eigenvectors corresponding to the largest d1 eigenvalues form the projection matrix W. We set a reconstruction threshold t from the perspective of reconstruction, also known as information cumulative contribution rate, and then select the minimum d1 value that makes the following formula valid:

≥ 𝑡
By retaining projection matrix W and sample mean vector  ̅ , samples can be projected to lowdimensional space through simple vector subtraction and matrix-vector multiplication.
Meanwhile, the samples in low-dimensional space can also be reconstructed in highdimensional space by discarding part of the information, as shown in Figure S8.
Coefficient of determination R 2 is used to judge the fitting results between the reconstructed force curve and the original force curve: where n is the total number of points on the curve,   is the fitting value of the sequence   and  ̅ is the mean of the sequence   .The closer R 2 is to 1, the better the fitting is.

Figure S18
The amount of time it takes for grip strength to drop to 50% of maximum grip strength.

Figure S19
The amount of time it takes for grip strength to drop to 60% of maximum grip strength.

Figure S20
The amount of time it takes for grip strength to drop to 70% of maximum grip strength.

Figure S21
Static fatigue rate derived from the grip strength curve.Static fatigue rate=1-AUC/HAUC.AUC is the integral of grip strength over the interval [Tmax, Ttotal].HAUC is equal to maximum grip strength times Ttotal-Tmax.Tmax is defined as the time when maximum grip strength occurs.Ttotal is defined as the total time of a single grasp.

Figure S1. a
Figure S1. a Representative volume element (RVEs) for the two metamaterials considered in this work.The solid line is MM-F voids, the dotted line is MM-C voids.b Schematic diagram of MM-F and MM-C samples.c Four-step mold casting procedure for the specimens.

Figure S3. a
Figure S3. a Experimental reaction force and open-circuit voltage of the MM-F and MM-C with the PVDF undergoing five cycles of loading and unloading.b The forces endured by the two metamaterials at c the minimum and maximum voltages output by the PVDF films.The hollow histograms refer to the data related to PVDF1 and the solid histograms represent the data related to PVDF2.Mean values are shown and error bars represent ±s.d.(n=4 samples per group), as analyzed by one-way ANOVA with post hoc t-tests with Bonferroni correction.

Figure S4 .
Figure S4.Voltage curves of PVDF1 and PVDF2 with time under 200 cycles of compression.b The maximum and minimum voltages of PVDF1 and PVDF2 with time under 200 cycles of compression.c Voltage curves of PVDF1 and PVDF2 over time for the last ten cycles of 200 compressions.

Figure S5 .
Figure S5.Mechanical and electrical responses for experiments under other 5 different compression amplitudes when compression speed is equal to 500 mm/min.
Figure S6. a The values of the maximum and minimum voltages for PVDF2 under different compression displacements and compression velocities.b The values of the maximum and minimum voltages for PVDF2 under three types of compression displacements and three types of compression velocities.Mean values are shown and error bars represent ±s.d.(n=16-40 samples per group), as analyzed by one-way ANOVA with post hoc t-tests with Bonferroni correction.

Figure S7
Figure S7 Schematic diagram of PVDF-induced charge variation during a grip measurement cycle.

Figure S8. a
Figure S8. a The linear correlation coefficient between target grip force and output target force.90%, 5% and 5% of the data sets are used for training, validating and testing, respectively.bThe absolute error between target grip force and output target force for three types of compression.

Figure S9. a
Figure S9. a The flow chat of machine learning used to map from voltage data to 15 main features of force data.b The flowchart about dimensionality reduction and reconstitution.

Figure S10. a Figure S11 .
Figure S10.a The linear correlation coefficient of training, validation, test and all datasets between the output (15 main features of force) of the neural network after training and the actual target output.70%, 15% and 15% of the data sets are used for training, validating and testing, respectively.b Coefficient of determination R 2 between the original force curve and the reconstructed force curve for all samples of PVDF1.

Figure S14 .
Figure S14.Training results for MM-F (PDMS 10:1).a Probability distribution of relative errors in maximum grip strength measurements.b The confusion matrix for recognition of three types of electromechanical classifications.c The confusion matrix for recognition of three types of compression speed.The recognition accuracies of training set (90%) and test set (10%) are both 100%.d Coefficient of determination R 2 between the original force curve and the reconstructed force curve for all samples.The recognition accuracies of training set (90%) and test set (10%) are 98.0% and 91.1%, respectively.

Figure S15 .
Figure S15.The fatigue rate derived from the grip strength curve.

Figure S16 .
Figure S16.The area under curve derived from the grip strength curve.

Figure S17 .
Figure S17.The grip velocity derived from the grip strength curve. 1 represents Slow, 2 represents Medium and 3 represents Fast.

Figure S25
Figure S25The grip velocity (a), the area under the curve (b), the static fatigue rate (c), the amount of time it takes for grip strength to drop to 50% (d),60% (e) and 70% (f) of maximum grip strength for two female schizophrenic patients.