In this paper, we measured the privacy budgets and utilities of different differentially private mechanisms combined with different machine learning models that forecast traffic congestion at future timestamps. We expected the ANNs combined with the Staircase mechanism to perform the best with every value in the privacy budget range, especially with the medium high values of the privacy budget. In this study, we used the Autoregressive Integrated Moving Average (ARIMA) and neural network models to forecast and then added differentially private Laplacian, Gaussian, and Staircase noise to our datasets. We tested two real traffic congestion datasets, experimented with the different models, and examined their utility for different privacy budgets. We found that a favorable combination for this application was neural networks with the Staircase mechanism. Our findings identify the optimal models when dealing with tricky time series forecasting and can be used in non-traffic applications like disease tracking and population growth.

In the battle against Alzheimer's disease, early detection is critical to mitigating symptoms in patients. Here, the authors use a collection of MRI scans, layering with deep learning computer modeling, to investigate early stages of AD which can be hard to catch by human eye. Their model is successful, able to outperform previous models, and detected regions of interest in the brain for further consideration.

Here, seeking to better understand the effects of gadolinium-based contrast agents, dyes typically used for MRI scans, the authors evaluated the activity of catalase found in beef liver both with and without gadodiamide when exposed to hydrogen peroxide. They found that gadioamide did not significantly inhibit catalase's activity, attributing this lack of effects to the chelating agent found in gadodiamide.

Van der Woude syndrome is a common birth defect caused by mutations in the gene Irf6. In this project, students used microarray expression analysis from wild-type and Irf6-deficient mice in order to identify gene networks or pathways differentially regulated due to the Irf6 mutation. They found NF-κB pathway to be activated in deficient mice.

As digital tools become more prevalent in medicine, the ability for individuals to understand and take actions based on what they read on the internet is crucial. eHealth literacy is defined as as the ability to seek, find, understand, and evaluate health information from electronic sources and apply the knowledge gained to addressing or solving a health problem. In general, Americans have low eHealth literacy rates. However, limited research has been conducted to understand the eHealth literacy level among older Chinese adult immigrants in the U.S. To determine the eHealth literacy of elderly Chinese immigrants, we sent out an eHealth survey and relevant computer skills survey using a modified version of the eHEALS (eHealth Literacy Scale) health literacy test. We hypothesized that elders who consumed more electronic health content would have a higher eHealth literacy score. The results of this survey showed that there was a positive correlation between the frequency of electronic health information consumption and the participant's eHealth literacy rate. In addition, the results of our computer literacy test show that the frequency of consumption and computer literacy are positively correlated as well. There is a strong positive correlation between the level of computer skills and eHealth literacy of participants. These results reveal possible steps individuals can take to reduce health misinformation and improve their own health by attaining, understanding, and taking action on health material on the internet.

This study's goal was to identify the Mach numbers for which electrostatic drag and heat transfer manipulation would be most applicable inside the stratosphere. The experiments were conducted using computational fluid dynamics software. The study demonstrated that, on average, higher Mach speeds resulted in a considerably higher potential decrease in density. The study highlights that further research on the surface charge method is warranted to explore higher hypersonic speeds within the stratosphere.

In this study, the authors determined whether tautomerization dynamics in protic and aprotic solvents displayed differences in reaction rates and in the proportion of the keto and enol tautomers present.

Irrespective of the final application of a molecule, synthetic accessibility is the rate-determining step in discovering and developing novel entities. However, synthetic complexity is challenging to quantify as a single metric, since it is a composite of several measurable metrics, some of which include cost, safety, and availability. Moreover, defining a single synthetic accessibility metric for both natural products and non-natural products poses yet another challenge given the structural distinctions between these two classes of compounds. Here, we propose a model for synthetic accessibility of all chemical compounds, inspired by the Central Limit Theorem, and devise a novel synthetic accessibility metric assessing the overall feasibility of making chemical compounds that has been fitted to a Gaussian distribution.

With advancements in machine learning a large data scale, high throughput virtual screening has become a more attractive method for screening drug candidates. This study compared the accuracy of molecular descriptors from two cheminformatics Mordred and PaDEL, software libraries, in characterizing the chemo-structural composition of 53 compounds from the non-nucleoside reverse transcriptase inhibitors (NNRTI) class. The classification model built with the filtered set of descriptors from Mordred was superior to the model using PaDEL descriptors. This approach can accelerate the identification of hit compounds and improve the efficiency of the drug discovery pipeline.

Modern artificial neural networks have been remarkably successful in various applications, from speech recognition to computer vision. However, it remains less clear whether they can implement abstract concepts, which are essential to generalization and understanding. To address this problem, the authors investigated the above vs. below task, a simple concept-based task that honeybees can solve, using a conventional neural network. They found that networks achieved 100% test accuracy when a visual target was presented below a black bar, however only 50% test accuracy when a visual target was presented below a reference shape.