Moving toward autonomous manufacturing by accelerating hydrodynamic fabrication of microstructures using deep neural networks

Objective: To investigate the influence of porosity and Young's modulus of gelatin methacrylate anhydride (GelMA) hydrogels on the biological behavior of mouse bone marrow mesenchymal stem cells (BMSCs), and to provide a theoretical basis for the development of tissue engineering scaffolds for skin wound repair. Methods: This study was an experimental study. GelMA hydrogels that were prepared by photo-crosslinking 50 g/L GelMA solution for 30 and 60 seconds and 150 g/L GelMA solution for 30 and 60 seconds, were designated respectively as low-concentration short-time group, low-concentration long-time group, high-concentration short-time group, and high-concentration long-time group. The porosity of the 4 groups of hydrogels after freeze-drying for 48 hours was measured and calculated using Image J 1.54 software. The Young's modulus of the 4 groups of hydrogels was detected using a Young's modulus tester. The correlation between the porosity and Young's modulus was evaluated and a predictive model between them was established. The remaining mass percentages of the 4 groups of hydrogels after soaking in phosphate buffered saline for 7 days were calculated. BMSCs were isolated from the femurs of 10 C57BL/6 mice aged 1 day of unknown sex. Cell-loaded GelMA hydrogels were prepared by mixing BMSCs with 50 g/L GelMA solution and photo-crosslinking for 30 and 60 seconds, and by mixing BMSCs with 150 g/L GelMA solution and photo-crosslinking for 30 and 60 seconds, and were designated respectively as low-concentration short-time cell-loaded group, low-concentration long-time cell-loaded group, high-concentration short-time cell-loaded group, and high-concentration long-time cell-loaded group. The cell survival in the 4 cell-loaded groups of hydrogels after 7 days of culture was detected using a live/dead cell kit, and the cell survival rate was calculated. After 10 days of osteogenic induction culture of the hydrogels in the 4 cell-loaded groups, the mRNA expressions of osteogenic-related factors Runt-related transcription factor 2 (Runx2) and alkaline phosphatase (ALP), and stemness-related factors SRY-box transcription factor 2 (Sox2) and Nanog in the cells were detected by real-time fluorescence quantitative reverse transcription polymerase chain reaction; the protein expressions of Runx2 and Nanog in the cells were detected by immunofluorescence method. Except for the protein expression (qualitative observation), the sample size of each group for the other indicators was three. Results: The porosity of hydrogels in low-concentration short-time group, low-concentration long-time group, high-concentration short-time group, and high-concentration long-time group after freeze-drying for 48 hours decreased successively, being (92.7±0.9)%, (85.0±1.8)%, (68.6±1.2)%, and (56.3±5.8)%, respectively; the Young's modulus increased successively, being (5.933±0.020), (7.803±0.089), (20.772±0.106), and (22.498±0.060) kPa, respectively. Except for the Young's modulus of hydrogels between the high-concentration long-time group and high-concentration short-time group (P>0.05), the pairwise comparisons of hydrogel porosity or Young's modulus among the remaining groups showed statistically significant differences (P<0.05). The porosity and Young's modulus of the hydrogels were significantly negatively correlated (R2=0.91, P<0.05), and a linear equation predictive model between them was established accordingly. After soaking for 7 days, the remaining mass percentages of hydrogels in low-concentration long-time group, high-concentration short-time group, and high-concentration long-time group were significantly higher than that in low-concentration short-time group (P<0.05), and the remaining mass percentage of hydrogels in high-concentration long-time group was significantly higher than that in low-concentration long-time group (P<0.05). After 7 days of culture, a large number of live cells was observed in the 4 cell-loaded groups of hydrogels, and there was no statistically significant difference in the overall comparison of cell survival rate among the groups (P>0.05). After 10 days of osteogenic induction culture, the mRNA expressions of Runx2 and ALP in the cells of hydrogels in high-concentration long-time cell-loaded group were significantly higher than those in the other three cell-loaded groups (P<0.05), and the mRNA expressions of Sox2 and Nanog in the cells of hydrogels in low-concentration short-time cell-loaded group were significantly higher than those in the other three cell-loaded groups (P<0.05); the protein expression of Runx2 in the cells of hydrogels in high-concentration long-time cell-loaded group was the highest, and the protein expression of Nanog in the cells of hydrogels in low-concentration short-time cell-loaded group was the highest. Conclusions: The porosity and Young's modulus of GelMA hydrogel synergistically regulate the biological behavior of mouse BMSCs, where high porosity and low Young's modulus are conducive to maintaining stemness, while low porosity and high Young's modulus drive osteogenic differentiation.

Understanding the mechanical and thermodynamic properties of transition-metal dichalcogenides (TMDs) and their heterostructures is pivotal for advancing the development of flexible semiconductor devices, and molecular dynamics (MD) simulation is widely applied to study these properties. However, current uncertainties persist regarding the efficacy of empirical potentials in MD simulations to accurately describe the intricate performance of complex interfaces within heterostructures. This study addresses these challenges by developing an interatomic potential based on deep neural networks and first-principles calculations. Specifically focusing on MoS2/WS2 heterostructures, our approach aims to predict Young's modulus and thermal conductivities. The potential's effectiveness is demonstrated through the validation of structural features, mechanical properties, and thermodynamic characteristics, revealing close alignment with values derived from first-principles calculations. A noteworthy finding is the substantial influence of the load direction on Young's modulus of heterostructures. Furthermore, our results highlight that the interfacial thermal conductance of the MoS2/WS2 heterostructures is considerably larger than that of graphene-based interfaces. The potential developed in this work facilitates large-scale material simulations, bridging the gap with first-principles calculations. Notably, it outperforms empirical potentials under interface conditions, establishing its significant competitiveness in simulation computations. Our approach not only contributes to a deeper understanding of TMDs and heterostructures but also presents a robust tool for the simulation of their mechanical and thermal behaviors, paving the way for advancements in flexible semiconductor device manufacturing.

A robust and extendable sheath flow interface with minimal dead volume for coupling CE with ESI-MS

Polydimethylsiloxane (PDMS) has been the pivotal materials for microfluidic technologies with tremendous amount of lab-on-a-chip devices made of PDMS microchannels. While molding-based soft-lithography approach has been extremely successful in preparing various PDMS constructs, some complex features have to been achieved through more complicated microfabrication techniques that involve dry etching of PDMS. Several recipes have been reported for reactive ion etching (RIE) of PDMS; however, the etch rates present large variations, even for the same etching recipe, which poses challenges in adopting this process for device fabrication. Through systematic characterization of the Young's modulus of PDMS films and RIE etch rate, we show that the etch rate is closely related to the polymer cross-link density in the PDMS with a higher etch rate for a lower PDMS Young's modulus. Our results could provide guidance to the fabrication of microfluidic devices involving dry etching of PDMS.

The mechanical/thermal/electrical properties on-demand design of CNTs-reinforced nanocomposites is a key scientific issue that limits the development of new-generation smart nanomaterials, and the establishment of a corresponding unified theoretical prediction model for the mechanical/thermal/electrical properties is the foundation of nanocomposites. Based on the equivalent medium theory (EMT) obtained by Maxwell far-field matching, a unified mechanical/thermal/electrical modified EMT model is established by introducing Young's modulus, thermal conductivity, and electrical conductivity to the thin filler-matrix's interlayer. According to literature, the proposed model was employed to theoretically calculate the variations in the overall Young's modulus, thermal conductivity, and electrical conductivity of CNTs-reinforced nanocomposites with respect to the volume concentration of CNT fillers. Then, the applicability of the proposed theoretical model was validated in comparison with the experimental measurements. Numerical calculations showed that the interface is a key factor affecting the mechanical/thermal/electrical properties of CNTs-reinforced nanocomposites, and strengthening the interfacial effect is an effective way to enhance the overall properties of nanocomposites. In addition, the aspect ratio of CNT fillers also significantly affects the material properties of the CNT fillers interface phase and the CNTs-reinforced nanocomposites. By fitting the experimental data, the calculation expressions of the aspect ratios of CNT fillers on the Young's modulus, thermal conductivity, and electrical conductivity of the CNT fillers interfacial phase are quantitatively given, respectively.

Today fused filament fabrication is one of the most widely used additive manufacturing techniques to manufacture high performance materials. This method entails a complexity associated with the selection of their appropriate manufacturing parameters. Due to the potential to replace poly‐ether‐ether‐ketone in many engineering components, polyphenylene sulfide (PPS) was selected in this study as a base material for 3D printing. Using central composite design and response surface methodology (RSM), nozzle temperature (T), printing speed (S), and layer thickness (L) were systematically studied to optimize the output responses namely Young's modulus, tensile strength, and degree of crystallinity. The results showed that the layer thickness was the most influential printing parameter on Young's modulus and degree of crystallinity. According to RSM, the optimum factor levels were achieved at 338°C nozzle temperature, 30 mm/s printing speed, and 0.17 mm layer thickness. The optimized post printed PPS parts were then annealed at various temperatures to erase thermal residual stress generated during the printing process and to improve the degree of crystallinity of printed PPS's parts. Results showed that annealing parts at 200°C for 1 hr improved significantly the thermal, structural, and tensile properties of printed PPS's parts.

This paper describes a highly controllable method to generate hollow calcium alginate microfibers using a double co-axial flow microdevice. The microdevice was fabricated by concentric assembly of two modules; each module consisted of a shortened cone-pulled glass capillary embedded in a polymethylmethacrylate fluidic block. Using this microdevice, cylindrical hollow calcium alginate microfibers with either straight or helical inner walls were stably and continuously generated. The radii of the hollow microfibers were not arbitrary, and in fact, the ratio of the outer to inner diameter was inversely correlated with the combination of core flow rate and the first sheath flow rate. The relationships between the geometrical features of the helix and the flow rates were also analyzed. The helical pitch and the spiral radius of the helical hollow microfibers were strongly influenced by the second sheath flow rate. Finally, guidelines for generating highly controllable straight and helical hollow microfibers and fabricating a seamless flow connector using this microfluidic device are suggested.

Coulter counters are analytical microfluidic instrument used to measure the size and concentration of biological cells or colloid particles suspended in electrolyte. The underlying working mechanism of Coulter counters is the Coulter principle which relies on the fact that when low-conductive cells pass through an electric field these cells cause disturbances in the measurement (current or voltage). Useful information about these cells can be obtained by analyzing these disturbances if an accurate correlation between the measured disturbances and cell characteristics. In this paper we use computational fluid dynamics method to investigate this correlation. The flow field is described by solving the Navier-Stokes equations, the electric field is represented by a Laplace’s equation in which the conductivity is calculated from the Navier-Stokes equations, and the cell motion is calculated by solving the equations of motion. The accuracy of the code is validated by comparing with analytical solutions. The study is based on a coplanar Coulter counter with three inlets that consist of two sheath flow inlet and one conductive flow inlet. The effects of diffusivity, cell size, sheath flow rate, and cell geometry are discussed in details. The impacts of electrode size, gap between electrodes and electrode location on the measured distribution are also studied.

Effects of swelling on the mechanical properties of a pH-sensitive hydrogel for use in microfluidic devices

In this article, Gallium nitride (GaN) cantilevers integrated with AlGaN/GaN high electron mobility transistors (HEMTs) are fabricated with a GaN-on-silicon platform, which is fully compatible with the standard HEMTs fabrication process. A type of micro-bending test was used to characterize the piezoresponse of AlGaN/GaN HEMT on the GaN cantilever. The modulation capability of the AlGaN/GaN heterostructures under different gate bias voltages for either high sensitivity or large output signals was demonstrated. A pulsed vibration measurement technique was used to evaluate the Young's modulus of the suspended GaN cantilevers, yielding a Young's modulus of ∼293 GPa.

Multizone pressure chemical mechanical planarization (CMP), which is regarded as irreplaceable in semiconductor manufacturing, is highly dependent on a process control strategy. However, predictive models with strict rules for advanced CMP multizone pressure process control have rarely been examined, even with the development of machine learning applications in semiconductor fabrication. Establishing a precise data-driven model that takes the fabrication conditions into consideration is urgent for controlling the CMP process. A novel structure for the CMP material removal rate model consisting of two parts is proposed and validated according to split pressure experiments and feature importance learning by this article. The first part was devoted to the prediction of the polishing time and time dependent problem with a deep neural network. The other part focused on the mutual effect of the pressure distribution on the multizone material removal rate using deep neural network with fully connected layers. To improve the prediction model of the first part, this study devises an advanced predictive model that combines a bidirectional gated recurrent unit network with a multi-head self-attention module and a one-dimensional neural network. This deep network extracts time-varying factors from industrial dataset and establish a nonlinear mapping function from CMP process features to material removal rate with no pressure variables. Both part of this structure is tested and evaluated for the feasibility of the predictive model using dataset from industrial fabrication.

Convolutional neural network-based wind pressure prediction on low-rise buildings

This paper covers predicting high-resolution electricity peak demand features given lower-resolution data. This is a relevant setup as it answers whether limited higher-resolution monitoring helps to estimate future high-resolution peak loads when the high-resolution data is no longer available. That question is particularly interesting for network operators considering replacing high-resolution monitoring by predictive models due to economic considerations. We propose models to predict half-hourly minima and maxima of high-resolution (every minute) electricity load data while model inputs are of a lower resolution (30 min). We combine predictions of generalized additive models (GAM) and deep artificial neural networks (DNN), which are popular in load forecasting. We extensively analyze the prediction models, including the input parameters’ importance, focusing on load, weather, and seasonal effects. The proposed method won a data competition organized by Western Power Distribution, a British distribution network operator. In addition, we provide a rigorous evaluation study that goes beyond the competition frame to analyze the models’ robustness. The results show that the proposed methods are superior to the competition benchmark concerning the out-of-sample root mean squared error (RMSE). This holds regarding the competition month and the supplementary evaluation study, which covers an additional eleven months. Overall, our proposed model combination reduces the out-of-sample RMSE by 57.4% compared to the benchmark.

In recent decades, the global incidence of dengue has increased. Affected countries have responded with more effective surveillance strategies to detect outbreaks early, monitor the trends, and implement prevention and control measures. We have applied newly developed machine learning approaches to identify laboratory-confirmed dengue cases from 4,894 emergency department patients with dengue-like illness (DLI) who received laboratory tests. Among them, 60.11% (2942 cases) were confirmed to have dengue. Using just four input variables [age, body temperature, white blood cells counts (WBCs) and platelets], not only the state-of-the-art deep neural network (DNN) prediction models but also the conventional decision tree (DT) and logistic regression (LR) models delivered performances with receiver operating characteristic (ROC) curves areas under curves (AUCs) of the ranging from 83.75% to 85.87% [for DT, DNN and LR: 84.60% ± 0.03%, 85.87% ± 0.54%, 83.75% ± 0.17%, respectively]. Subgroup analyses found all the models were very sensitive particularly in the pre-epidemic period. Pre-peak sensitivities (<35 weeks) were 92.6%, 92.9%, and 93.1% in DT, DNN, and LR respectively. Adjusted odds ratios examined with LR for low WBCs [≤ 3.2 (x103/μL)], fever (≥38°C), low platelet counts [< 100 (x103/μL)], and elderly (≥ 65 years) were 5.17 [95% confidence interval (CI): 3.96–6.76], 3.17 [95%CI: 2.74–3.66], 3.10 [95%CI: 2.44–3.94], and 1.77 [95%CI: 1.50–2.10], respectively. Our prediction models can readily be used in resource-poor countries where viral/serologic tests are inconvenient and can also be applied for real-time syndromic surveillance to monitor trends of dengue cases and even be integrated with mosquito/environment surveillance for early warning and immediate prevention/control measures. In other words, a local community hospital/clinic with an instrument of complete blood counts (including platelets) can provide a sentinel screening during outbreaks. In conclusion, the machine learning approach can facilitate medical and public health efforts to minimize the health threat of dengue epidemics. However, laboratory confirmation remains the primary goal of surveillance and outbreak investigation.

Introduction: There has been little study regarding prediction of acute exacerbation. Aims and Objective: This study aimed to develop prediction model of COPD acute exacerbation with big data by machine learning methods. Methods: We investigated data from 594 COPD patients who were enrolled in the KOCOSS cohort. Smoking status, lung function, body mass index, and COPD assessment test (CAT) score were collected from cohort data. We merged patients’ information with the Korean Health Insurance database. Comorbidity, health care utilization, moderate to severe exacerbation, and COPD medications between 2008 and 2012 were collected. We also collected daily air pollution level, temperature, humidity, and wind velocity. Data on the activities of respiratory viruses were collected. Prediction model was developed by deep learning methods and also statistical method. Deep neural network (DNN), random forest, and generalized estimating equation (GEE) were utilized. Results: Area under the curve (AUC) value for prediction of acute exacerbation was highest by random forest method (0.8722) followed by GEE (0.8617) and DNN (0.8493). Sensitivity was highest by random forest (0.8613) followed by DNN (0.8000) and GEE (0.7613). In GEE analysis, female, CAT score, FEV1 (%), number of exacerbations in previous one year, use of bronchodilator, history of asthma, influenza, and human coronavirus were significantly associated with acute exacerbation. Conclusions: Prediction model for acute exacerbation of COPD was developed with big data by machine learning methods. AUC and sensitivity were higher in model by machine learning methods compared with GEE.