Search

Background Fatty liver disease (FLD) arises from the accumulation of fat in the liver and may cause liver inflammation, which, if not well controlled, may develop into liver fibrosis, cirrhosis, or even hepatocellular carcinoma. Objective We describe the construction of machine-learning models for current-visit prediction (CVP), which can help physicians obtain more information for accurate diagnosis, and next-visit prediction (NVP), which can help physicians provide potential high-risk patients with advice to effectively prevent FLD. Methods The large-scale and high-dimensional dataset used in this study comes from Taipei MJ Health Research Foundation in Taiwan. We used one-pass ranking and sequential forward selection (SFS) for feature selection in FLD prediction. For CVP, we explored multiple models, including k-nearest-neighbor classifier (KNNC), Adaboost, support vector machine (SVM), logistic regression (LR), random forest (RF), Gaussian naïve Bayes (GNB), decision trees C4.5 (C4.5), and classification and regression trees (CART). For NVP, we used long short-term memory (LSTM) and several of its variants as sequence classifiers that use various input sets for prediction. Model performance was evaluated based on two criteria: the accuracy of the test set and the intersection over union/coverage between the features selected by one-pass ranking/SFS and by domain experts. The accuracy, precision, recall, F-measure, and area under the receiver operating characteristic curve were calculated for both CVP and NVP for males and females, respectively. Results After data cleaning, the dataset included 34,856 and 31,394 unique visits respectively for males and females for the period 2009-2016. The test accuracy of CVP using KNNC, Adaboost, SVM, LR, RF, GNB, C4.5, and CART was respectively 84.28%, 83.84%, 82.22%, 82.21%, 76.03%, 75.78%, and 75.53%. The test accuracy of NVP using LSTM, bidirectional LSTM (biLSTM), Stack-LSTM, Stack-biLSTM, and Attention-LSTM was respectively 76.54%, 76.66%, 77.23%, 76.84%, and 77.31% for fixed-interval features, and was 79.29%, 79.12%, 79.32%, 79.29%, and 78.36%, respectively, for variable-interval features. Conclusions This study explored a large-scale FLD dataset with high dimensionality. We developed FLD prediction models for CVP and NVP. We also implemented efficient feature selection schemes for current- and next-visit prediction to compare the automatically selected features with expert-selected features. In particular, NVP emerged as more valuable from the viewpoint of preventive medicine. For NVP, we propose use of feature set 2 (with variable intervals), which is more compact and flexible. We have also tested several variants of LSTM in combination with two feature sets to identify the best match for male and female FLD prediction. More specifically, the best model for males was Stack-LSTM using feature set 2 (with 79.32% accuracy), whereas the best model for females was LSTM using feature set 1 (with 81.90% accuracy).

BACKGROUND Fatty liver disease (FLD) arises from the accumulation of fat in the liver and may cause liver inflammation, which, if not well controlled, may develop into liver fibrosis, cirrhosis, or even hepatocellular carcinoma. OBJECTIVE We describe the construction of machine-learning models for current-visit prediction (CVP), which can help physicians obtain more information for accurate diagnosis, and next-visit prediction (NVP), which can help physicians provide potential high-risk patients with advice to effectively prevent FLD. METHODS The large-scale and high-dimensional dataset used in this study comes from Taipei MJ Health Research Foundation in Taiwan. We used one-pass ranking and sequential forward selection (SFS) for feature selection in FLD prediction. For CVP, we explored multiple models, including k-nearest-neighbor classifier (KNNC), Adaboost, support vector machine (SVM), logistic regression (LR), random forest (RF), Gaussian naïve Bayes (GNB), decision trees C4.5 (C4.5), and classification and regression trees (CART). For NVP, we used long short-term memory (LSTM) and several of its variants as sequence classifiers that use various input sets for prediction. Model performance was evaluated based on two criteria: the accuracy of the test set and the intersection over union/coverage between the features selected by one-pass ranking/SFS and by domain experts. The accuracy, precision, recall, F-measure, and area under the receiver operating characteristic curve were calculated for both CVP and NVP for males and females, respectively. RESULTS After data cleaning, the dataset included 34,856 and 31,394 unique visits respectively for males and females for the period 2009-2016. The test accuracy of CVP using KNNC, Adaboost, SVM, LR, RF, GNB, C4.5, and CART was respectively 84.28%, 83.84%, 82.22%, 82.21%, 76.03%, 75.78%, and 75.53%. The test accuracy of NVP using LSTM, bidirectional LSTM (biLSTM), Stack-LSTM, Stack-biLSTM, and Attention-LSTM was respectively 76.54%, 76.66%, 77.23%, 76.84%, and 77.31% for fixed-interval features, and was 79.29%, 79.12%, 79.32%, 79.29%, and 78.36%, respectively, for variable-interval features. CONCLUSIONS This study explored a large-scale FLD dataset with high dimensionality. We developed FLD prediction models for CVP and NVP. We also implemented efficient feature selection schemes for current- and next-visit prediction to compare the automatically selected features with expert-selected features. In particular, NVP emerged as more valuable from the viewpoint of preventive medicine. For NVP, we propose use of feature set 2 (with variable intervals), which is more compact and flexible. We have also tested several variants of LSTM in combination with two feature sets to identify the best match for male and female FLD prediction. More specifically, the best model for males was Stack-LSTM using feature set 2 (with 79.32% accuracy), whereas the best model for females was LSTM using feature set 1 (with 81.90% accuracy).

Formation of precise and high-resolution silica micropatterns on polymer substrates is of importance in surface structuring for flexible device fabrication of optics, microelectronic, and biotechnology. To achieve that, substrates modified with affinity-patterns serve as a strategy for site-selective deposition. In the present paper, vacuum ultraviolet (VUV) treatment is utilized to achieve spatially-controlled surface functionalization on a cyclo-olefin polymer (COP) substrate. An organosilane, 2, 4, 6, 8-tetramethylcyclotetrasiloxane (TMCTS), preferentially deposits on the functionalized regions. Well-defined patterns of TMCTS are formed with a minimum feature of ∼500 nm. The secondary VUV/(O)-treatment converts TMCTS into SiOx, meanwhile etches the bare COP surface, forming patterned SiOx/COP microstructures with an average height of ∼150 nm. The resulting SiOx patterns retain a good copy of TMCTS patterns, which are also consistent with the patterns of photomask used in polymer affinity-patterning. The high quality SiOx patterns are of interests in microdevice fabrication, and the hydrophilicity contrast and adjustable heights reveal their potential application as a “stamp” for microcontact printing (μCP) techniques.

This research investigated the risks involved in bicycle riding while using various sensory modalities to deliver training information. To understand the risks associated with using bike computers, this study evaluated hazard perception performance through lab-based simulations of authentic riding conditions. Analysing hazard sensitivity (d') of signal detection theory, the rider's response time, and eye glances provided insights into the risks of using bike computers. In this study, 30 participants were tested with eight hazard perception tasks while they maintained a cadence of 60 ± 5 RPM and used bike computers with different sensory displays, namely visual, auditory, and tactile feedback signals. The results indicated that synchronously using different sense organs to receive cadence feedback significantly affects hazard perception performance; direct visual information leads to the worst rider distraction, with a mean sensitivity to hazards (d') of −1.03. For systems with multiple interacting sensory aids, auditory aids were found to result in the greatest reduction in sensitivity to hazards (d' mean = −0.57), whereas tactile sensory aids reduced the degree of rider distraction (d' mean = −0.23). Our work complements existing work in this domain by advancing the understanding of how to design devices that deliver information subtly, thereby preventing disruption of a rider's perception of road hazards.

This study evaluated the impact of multi-sensory information cues from on-bicycle rider information assistance devices (OBRAD) on hazard perception performance. Experiments tested the impact of distraction from different combinations of visual, auditory and tactile sensory aids on the subject’s ability to maintain peddling frequency while conducting eight different tasks. The results indicate that the integrated use of different sensory cues (e.g., text, audible alerts and vibration) can decrease cognitive loading, with each sensory combination, particularly those involving tactile stimulation, having different levels of effect. Tactile sensory aids helped reduce the degree of rider distraction, thus helping maintain a high sensitivity to danger (hit rate mean: 0.34). Cycling performance was further improved through combining tactile stimuli with auditory cues for assistance in the secondary task. The implications of these findings and the need to integrate and manage complex OBRAD information systems are discussed.

The proximity of cells in three-dimensional (3D) organization maximizes the cell-cell communication and signaling that are critical for cell function. In this study, 3D cell aggregates composed of human umbilical vein endothelial cells (HUVECs) and cord-blood mesenchymal stem cells (cbMSCs) were used for therapeutic neovascularization to rescue tissues from critical limb ischemia. Within the cell aggregates, homogeneously mixed HUVECs and cbMSCs had direct cell-cell contact with expressions of endogenous extracellular matrices and adhesion molecules. Although dissociated HUVECs/cbMSCs initially formed tubular structures on Matrigel, the grown tubular network substantially regressed over time. Conversely, 3D HUVEC/cbMSC aggregates seeded on Matrigel exhibited an extensive tubular network that continued to expand without regression. Immunostaining experiments show that, by differentiating into smooth muscle cell (SMC) lineages, the cbMSCs stabilize the HUVEC-derived tubular network. The real-time PCR analysis results suggest that, through myocardin, TGF-β signaling regulates the differentiation of cbMSCs into SMCs. Transplantation of 3D HUVEC/cbMSC aggregates recovered blood perfusion in a mouse model of hindlimb ischemia more effectively compared to their dissociated counterparts. The experimental results confirm that the transplanted 3D HUVEC/cbMSC aggregates enhanced functional vessel formation within the ischemic limb and protected it from degeneration. The 3D HUVEC/cbMSC aggregates can therefore facilitate the cell-based therapeutic strategies for modulating postnatal neovascularization.

Myocardial infarction is often associated with abnormalities in electrical function due to a massive loss of functioning cardiomyocytes. This work develops a mesh, consisting of aligned composite nanofibers of polyaniline (PANI) and poly(lactic-co-glycolic acid) (PLGA), as an electrically active scaffold for coordinating the beatings of the cultured cardiomyocytes synchronously. Following doping by HCl, the electrospun fibers could be transformed into a conductive form carrying positive charges, which could then attract negatively charged adhesive proteins (i.e. fibronectin and laminin) and enhance cell adhesion. During incubation, the adhered cardiomyocytes became associated with each other and formed isolated cell clusters; the cells within each cluster elongated and aligned their morphology along the major axis of the fibrous mesh. After culture, expression of the gap-junction protein connexin 43 was clearly observed intercellularly in isolated clusters. All of the cardiomyocytes within each cluster beat synchronously, implying that the coupling between the cells was fully developed. Additionally, the beating rates among these isolated cell clusters could be synchronized via an electrical stimulation designed to imitate that generated in a native heart. Importantly, improving the impaired heart function depends on electrical coupling between the engrafted cells and the host myocardium to ensure their synchronized beating.