Bike-Sharing Demand Prediction at Community Level under COVID-19 Using Deep Learning.

An important question in planning and designing bike-sharing services is to support the user's travel demand by allocating bikes at the stations in an efficient and reliable manner which may require accurate short-time demand prediction. This study focuses on the short-term forecasting, 15 min ahead, of the shared bikes demand in Montreal using a deep learning approach. Having a set of bike trips, the study first identifies 6 communities in the bike-sharing network using the Louvain algorithm. Then, four groups of LSTM-based architectures are adopted to predict pickup demand in each community. A univariate ARIMA model is also used to compare results as a benchmark. The historical trip data from 2017 to 2021 are used in addition to the extra inputs of demand related engineered features, weather conditions, and temporal variables. The selected timespan allows predicting bike demand during the COVID-19 pandemic. Results show that the deep learning models significantly outperform the ARIMA one. The hybrid CNN-LSTM achieves the highest prediction accuracy. Furthermore, adding the extra variables improves the model performance regardless of its architecture. Thus, using the hybrid structure enriched with additional input features provides a better insight into the bike demand patterns, in support of bike-sharing operational management.

Multi-Step Hourly Power Consumption Forecasting in a Healthcare Building with Recurrent Neural Networks and Empirical Mode Decomposition.

Short-term forecasting of electric energy consumption has become a critical issue for companies selling and buying electricity because of the fluctuating and rising trend of its price. Forecasting tools based on Artificial Intelligence have proved to provide accurate and reliable prediction, especially Neural Networks, which have been widely used and have become one of the preferred ones. In this work, two of them, Long Short-Term Memories and Gated Recurrent Units, have been used along with a preprocessing algorithm, the Empirical Mode Decomposition, to make up a hybrid model to predict the following 24 hourly consumptions (a whole day ahead) of a hospital. Two different datasets have been used to forecast them: a univariate one in which only consumptions are used and a multivariate one in which other three variables (reactive consumption, temperature, and humidity) have been also used. The results achieved show that the best performances were obtained with the multivariate dataset. In this scenario, the hybrid models (neural network with preprocessing) clearly outperformed the simple ones (only the neural network). Both neural models provided similar performances in all cases. The best results (Mean Absolute Percentage Error: 3.51% and Root Mean Square Error: 55.06) were obtained with the Long Short-Term Memory with preprocessing with the multivariate dataset.

A model based study on the dynamics of COVID-19: Prediction and control.

As there is no vaccination and proper medicine for treatment, the recent pandemic caused by COVID-19 has drawn attention to the strategies of quarantine and other governmental measures, like lockdown, media coverage on social isolation, and improvement of public hygiene, etc to control the disease. The mathematical model can help when these intervention measures are the best strategies for disease control as well as how they might affect the disease dynamics. Motivated by this, in this article, we have formulated a mathematical model introducing a quarantine class and governmental intervention measures to mitigate disease transmission. We study a thorough dynamical behavior of the model in terms of the basic reproduction number. Further, we perform the sensitivity analysis of the essential reproduction number and found that reducing the contact of exposed and susceptible humans is the most critical factor in achieving disease control. To lessen the infected individuals as well as to minimize the cost of implementing government control measures, we formulate an optimal control problem, and optimal control is determined. Finally, we forecast a short-term trend of COVID-19 for the three highly affected states, Maharashtra, Delhi, and Tamil Nadu, in India, and it suggests that the first two states need further monitoring of control measures to reduce the contact of exposed and susceptible humans.

Utility of Big Data in Predicting Short-Term Blood Glucose Levels in Type 1 Diabetes Mellitus Through Machine Learning Techniques.

Machine learning techniques combined with wearable electronics can deliver accurate short-term blood glucose level prediction models. These models can learn personalized glucose-insulin dynamics based on the sensor data collected by monitoring several aspects of the physiological condition and daily activity of an individual. Until now, the prevalent approach for developing data-driven prediction models was to collect as much data as possible to help physicians and patients optimally adjust therapy. The objective of this work was to investigate the minimum data variety, volume, and velocity required to create accurate person-centric short-term prediction models. We developed a series of these models using different machine learning time series forecasting techniques suitable for execution within a wearable processor. We conducted an extensive passive patient monitoring study in real-world conditions to build an appropriate data set. The study involved a subset of type 1 diabetic subjects wearing a flash glucose monitoring system. We comparatively and quantitatively evaluated the performance of the developed data-driven prediction models and the corresponding machine learning techniques. Our results indicate that very accurate short-term prediction can be achieved by only monitoring interstitial glucose data over a very short time period and using a low sampling frequency. The models developed can predict glucose levels within a 15-min horizon with an average error as low as 15.43 mg/dL using only 24 historic values collected within a period of sex hours, and by increasing the sampling frequency to include 72 values, the average error is reduced to 10.15 mg/dL. Our prediction models are suitable for execution within a wearable device, requiring the minimum hardware requirements while at simultaneously achieving very high prediction accuracy.

Short-term prediction of coronary heart disease mortality in the Czech Republic based on data from 1968-2014.

OBJECTIVES: The aim was to explore the patterns of the coronary heart disease (CHD) mortality rates over the past almost 50 years (1968-2014) in the Czech Republic, and to predict the mortality rates in 2015-2019. METHODS: The number of deaths from CHD and the population size were stratified by sex and age. The mortality rates were age-standardized to European population. Their values in 2015-2019 were estimated using the joinpoint log-linear regression, local log-linear regression and negative binomial log-linear regression, separately for males and females. RESULTS: A positive change in the trend of the age-standardized mortality rates from CHD was detected after the collapse of communism in 1989. In 1991-2000, the mortality trend was sharply downward, with an annual percent change of -5.8 % for males and -5.2 % for females. In 2000-2014, the decreasing trend was not so sharp (-1.3 % for males and -0.7% for females), yet it should continue in 2015-2019. The crude mortality rates for females are slightly higher than those for males since 2007, however, they are increasing for both sexes. The mortality rates are rising mainly in the age group of 85+ years (in 2014, 25.4% of CHD deaths of males and 54.4% of females occurred at the age of 85+ years). CONCLUSIONS: The age-standardized mortality rates are predicted to decrease in 2015-2019, but the crude mortality rates should increase due to increase in average life expectancy. The burden of deaths is moving to the age group of 85 years and older, mainly in females. A total of 26,039 CHD deaths were registered in the Czech Republic in 2014, and 29,653 are predicted for 2019, if the current trends continue.

Improved Short-Term Clock Prediction Method for Real-Time Positioning.

The application of real-time precise point positioning (PPP) requires real-time precise orbit and clock products that should be predicted within a short time to compensate for the communication delay or data gap. Unlike orbit correction, clock correction is difficult to model and predict. The widely used linear model hardly fits long periodic trends with a small data set and exhibits significant accuracy degradation in real-time prediction when a large data set is used. This study proposes a new prediction model for maintaining short-term satellite clocks to meet the high-precision requirements of real-time clocks and provide clock extrapolation without interrupting the real-time data stream. Fast Fourier transform (FFT) is used to analyze the linear prediction residuals of real-time clocks. The periodic terms obtained through FFT are adopted in the sliding window prediction to achieve a significant improvement in short-term prediction accuracy. This study also analyzes and compares the accuracy of short-term forecasts (less than 3 h) by using different length observations. Experimental results obtained from International GNSS Service (IGS) final products and our own real-time clocks show that the 3-h prediction accuracy is better than 0.85 ns. The new model can replace IGS ultra-rapid products in the application of real-time PPP. It is also found that there is a positive correlation between the prediction accuracy and the short-term stability of on-board clocks. Compared with the accuracy of the traditional linear model, the accuracy of the static PPP using the new model of the 2-h prediction clock in N, E, and U directions is improved by about 50%. Furthermore, the static PPP accuracy of 2-h clock products is better than 0.1 m. When an interruption occurs in the real-time model, the accuracy of the kinematic PPP solution using 1-h clock prediction product is better than 0.2 m, without significant accuracy degradation. This model is of practical significance because it solves the problems of interruption and delay in data broadcast in real-time clock estimation and can meet the requirements of real-time PPP.

Predicting mine water inflow volumes using a decomposition-optimization algorithm-machine learning approach.

Disasters caused by mine water inflows significantly threaten the safety of coal mining operations. Deep mining complicates the acquisition of hydrogeological parameters, the mechanics of water inrush, and the prediction of sudden changes in mine water inflow. Traditional models and singular machine learning approaches often fail to accurately forecast abrupt shifts in mine water inflows. This study introduces a novel coupled decomposition-optimization-deep learning model that integrates Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), Northern Goshawk Optimization (NGO), and Long Short-Term Memory (LSTM) networks. We evaluate three types of mine water inflow forecasting methods: a singular time series prediction model, a decomposition-prediction coupled model, and a decomposition-optimization-prediction coupled model, assessing their ability to capture sudden changes in data trends and their prediction accuracy. Results show that the singular prediction model is optimal with a sliding input step of 3 and a maximum of 400 epochs. Compared to the CEEMDAN-LSTM model, the CEEMDAN-NGO-LSTM model demonstrates superior performance in predicting local extreme shifts in mine water inflow volumes. Specifically, the CEEMDAN-NGO-LSTM model achieves scores of 96.578 in MAE, 1.471% in MAPE, 122.143 in RMSE, and 0.958 in NSE, representing average performance improvements of 44.950% and 19.400% over the LSTM model and CEEMDAN-LSTM model, respectively. Additionally, this model provides the most accurate predictions of mine water inflow volumes over the next five days. Therefore, the decomposition-optimization-prediction coupled model presents a novel technical solution for the safety monitoring of smart mines, offering significant theoretical and practical value for ensuring safe mining operations.

Modified artificial neural network based on developed snake optimization algorithm for short-term price prediction.

Short-term prices prediction is a crucial task for participants in the electricity market, as it enables them to optimize their bidding strategies and mitigate risks. However, the price signal is subject to various factors, including supply, demand, weather conditions, and renewable energy sources, resulting in high volatility and nonlinearity. In this study, a novel approach is introduced that combines Artificial Neural Networks (ANN) with a newly developed Snake Optimization Algorithm (SOA) to forecast short-term price signals in the Nord Pool market. The snake optimization algorithm is utilized to optimize both the structure and weights of the neural network, as well as to select relevant input data based on the similarity of price curves and wind production. To evaluate the effectiveness of the proposed technique, experiments have been conducted using data from two regions of the Nord Pool market, namely DK-1 and SE-1, across different seasons and time horizons. The results demonstrate that the proposed technique surpasses two alternative methods based on Particle Swarm Optimization (PSO) and Genetic Algorithms-based Neural Network (PSOGANN) and Gravitational Search Optimization Algorithm-based Neural Network (GSONN), exhibiting superior accuracy and minimal error rates in short-term price prediction. The results show that the average MAPE index of the proposed technique for the DK-1 region is 3.1292%, which is 32.5% lower than the PSOGA method and 47.1% lower than the GSONN method. For the SE-1 region, the average MAPE index of the proposed technique is 2.7621%, which is 40.4% lower than the PSOGA method and 64.7% lower than the GSONN method. Consequently, the proposed technique holds significant potential as a valuable tool for market participants to enhance their decision-making and planning activities.

Research on renewable energy prediction technology: empirical analysis for Argentina and China.

Our world needs to develop clean energy to reach the target of carbon peak and carbon neutralization. As one of clean energy, wind energy should contribute to energy conservation and emission reduction. Wind power generation is an important field of wind energy application. However, the fluctuation and intermittency of wind can affect the safety of power system. Therefore, prediction of wind power accurately for wind power safety, dispatching, and power grid development is significant. This paper proposes a prediction model of wind power, and predicts the wind power of two wind farms. For the complex wind speed series, the variational modal decomposition (VMD) method is used to reduce its volatility before prediction. And this paper presents an improved method to improve the prediction efficiency when least square support vector machine (LSSVM) predicts stationary series. The prediction result shows that the proposed model improves the prediction of wind power effectively, provides an effective method for wind farm to predict the wind power, and makes contributions to reducing carbon emissions and environmental protection.

A time series based machine learning strategy for wastewater-based forecasting and nowcasting of COVID-19 dynamics.

Monitoring COVID-19 infection cases has been a singular focus of many policy makers and communities. However, direct monitoring through testing has become more onerous for a number of reasons, such as costs, delays, and personal choices. Wastewater-based epidemiology (WBE) has emerged as a viable tool for monitoring disease prevalence and dynamics to supplement direct monitoring. The objective of this study is to intelligently incorporate WBE information to nowcast and forecast new weekly COVID-19 cases and to assess the efficacy of such WBE information for these tasks in an interpretable manner. The methodology consists of a time-series based machine learning (TSML) strategy that can extract deeper knowledge and insights from temporal structured WBE data in the presence of other relevant temporal variables, such as minimum ambient temperature and water temperature, to boost the capability for predicting new weekly COVID-19 case numbers. The results confirm that feature engineering and machine learning can be utilized to enhance the performance and interpretability of WBE for COVID-19 monitoring, along with identifying the different recommended features to be applied for short-term and long-term nowcasting and short-term and long-term forecasting. The conclusion of this research is that the proposed time-series ML methodology performs as well, and sometimes better, than simple predictions that assume available and accurate COVID-19 case numbers from extensive monitoring and testing. Overall, this paper provides an insight into the prospects of machine learning based WBE to the researchers and decision-makers as well as public health practitioners for predicting and preparing the next wave of COVID-19 or the next pandemic.