Predictive Approach Research Articles

A hybrid machine learning approach for congestion prediction and warning

A hybrid machine learning approach for congestion prediction and warning

Electrical Impedance Tomography, Artificial Intelligence, and Variable Ventilation: Transforming Respiratory Monitoring and Treatment in Critical Care

Background: Electrical Impedance Tomography (EIT), combined with variable ventilation strategies and Artificial Intelligence (AI), is poised to revolutionize critical care by transitioning from reactive to predictive approaches. This integration aims to enhance patient outcomes through personalized interventions and real-time monitoring. Methods: this narrative review explores the principles and applications of EIT, variable ventilation, and AI in critical care. EIT impedance sensing creates dynamic images of internal physiology, aiding the management of conditions like Acute Respiratory Distress Syndrome (ARDS). Variable ventilation mimics natural breathing variability to improve lung function and minimize ventilator-induced lung injury. AI enhances EIT through advanced image reconstruction techniques, neural networks, and digital twin technology, offering more accurate diagnostics and tailored therapeutic interventions. Conclusions: the confluence of EIT, variable ventilation, and AI represents a significant advancement in critical care, enabling a predictive, personalized approach. EIT provides real-time insights into lung function, guiding precise ventilation adjustments and therapeutic interventions. AI integration enhances EIT diagnostic capabilities, facilitating the development of personalized treatment plans. This synergy fosters interdisciplinary collaborations and sets the stage for innovative research, ultimately improving patient outcomes and advancing the future of critical care.

Innovative approaches for accurate ozone prediction and health risk analysis in South Korea: The combined effectiveness of deep learning and AirQ+

Short-term exposure to ground-level ozone (O3) poses significant health risks, particularly respiratory and cardiovascular diseases, and mortality. This study addresses the pressing need for accurate O3 forecasting to mitigate these risks, focusing on South Korea. We introduce Deep Bias Correction (Deep-BC), a novel framework leveraging Convolutional Neural Networks (CNNs), to refine hourly O3 forecasts from the Community Multiscale Air Quality (CMAQ) model. Our approach involves training Deep-BC using data from 2016 to 2019, including CMAQ's 72-hour O3 forecasts, 31 meteorological variables from the Weather Research and Forecasting (WRF) model, and previous days' station measurements of 6 air pollutants. Deep-BC significantly outperforms CMAQ in 2021, reducing biases in O3 forecasts. Furthermore, we utilize Deep-BC's daily maximum 8-hour average O3 (MDA8 O3) forecasts as input for the AirQ+ model to assess O3's potential impact on mortality across seven major provinces of South Korea: Seoul, Busan, Daegu, Incheon, Daejeon, Ulsan, and Sejong. Short-term O3 exposure is associated with 0.40 % to 0.48 % of natural cause and respiratory deaths and 0.67 % to 0.81 % of cardiovascular deaths. Gender-specific analysis reveals higher mortality rates among men, particularly from respiratory causes. Our findings underscore the critical need for region-specific interventions to address air pollution's detrimental effects on public health in South Korea. By providing improved O3 predictions and quantifying its impact on mortality, this research offers valuable insights for formulating targeted strategies to mitigate air pollution's adverse effects. Moreover, we highlight the urgency of proactive measures in health policies, emphasizing the significance of accurate forecasting and effective interventions to safeguard public health from the deleterious effects of air pollution.

Analysis of intrinsic factors in accurate wave height prediction based on model interpretability

This study addresses the ‘black box issue’ in machine learning by utilizing interpretable model mechanisms for wave forecasting. The Shapley Additive Explanation (SHAP) method is integrated into various deep learning models to evaluate the contributions of different physical features to wave height prediction. By categorizing these features based on their importance levels, the primary factors influencing wave forecasting are thoroughly investigated. The results indicate that variations in feature contribution scores are linked to physical correlations, temporal correlations, model characteristics, and data quality. Remarkably, features directly related to fundamental physical processes, such as significant wave height and average wave period, are essential for accurate predictions. Different models display varying sensitivities to specific features over time, impacting prediction accuracy. This comprehensive analysis leads to an optimized prediction approach, significantly enhancing prediction efficiency and providing a solution for training deep learning models with limited training data.

Addressing data limitations in seizure prediction through transfer learning

According to the literature, seizure prediction models should be developed following a patient-specific approach. However, seizures are usually very rare events, meaning the number of events that may be used to optimise seizure prediction approaches is limited. To overcome such constraint, we analysed the possibility of using data from patients from an external database to improve patient-specific seizure prediction models. We present seizure prediction models trained using a transfer learning procedure. We trained a deep convolutional autoencoder using electroencephalogram data from 41 patients collected from the EPILEPSIAE database. Then, a bidirectional long short-term memory and a classifier layers were added on the top of the encoder part and were optimised for 24 patients from the Universitätsklinikum Freiburg individually. The encoder was used as a feature extraction module. Therefore, its weights were not changed during the patient-specific training. Experimental results showed that seizure prediction models optimised using pretrained weights present about four times fewer false alarms while maintaining the same ability to predict seizures and achieved more 13% validated patients. Therefore, results evidenced that the optimisation using transfer learning was more stable and faster, saving computational resources. In summary, adopting transfer learning for seizure prediction models represents a significant advancement. It addresses the data limitation seen in the seizure prediction field and offers more efficient and stable training, conserving computational resources. Additionally, despite the compact size, transfer learning allows to easily share data knowledge due to fewer ethical restrictions and lower storage requirements. The convolutional autoencoder developed in this study will be shared with the scientific community, promoting further research.

A modified mechanistic approach for predicting ribbon solid fraction at different roller compaction speeds

This research investigates the modeling of the pharmaceutical roller compaction process, focusing on the application of the Johanson model and the impact of varying roll speeds from 1 to 15 RPM on predictive accuracy of ribbon solid fraction. The classical Johanson’s model was integrated with a dwell time parameter leading to an expression of a floating correction factor as a function of roll speed. Through systematic analysis of the effect of different roll speeds on the solid fraction of ribbons composed of microcrystalline cellulose, lactose, and their blends, corrective adjustment to the Johanson model was found to depend on both roll speed and formulation composition. Interestingly, the correction factor demonstrated excellent correlation with the blend’s mechanical properties, namely yield stress (Py) and elastic modulus (E0), representative of the deformability of the powder. Validated by a multicomponent drug formulation with ±0.4–1.3 % differences, the findings underscore the utility of this modified mechanistic approach for precise prediction of ribbon solid fraction when Py or E0 is known for a given blend. Hence, this work advances the field by offering early insights for more accurate and controllable roller compaction operations during late-stage pharmaceutical manufacturing.

Insights into Parkinson’s Disease-Related Freezing of Gait Detection and Prediction Approaches: A Meta Analysis

Parkinson’s Disease (PD) is a complex neurodegenerative disorder characterized by a spectrum of motor and non-motor symptoms, prominently featuring the freezing of gait (FOG), which significantly impairs patients’ quality of life. Despite extensive research, the precise mechanisms underlying FOG remain elusive, posing challenges for effective management and treatment. This paper presents a comprehensive meta-analysis of FOG prediction and detection methodologies, with a focus on the integration of wearable sensor technology and machine learning (ML) approaches. Through an exhaustive review of the literature, this study identifies key trends, datasets, preprocessing techniques, feature extraction methods, evaluation metrics, and comparative analyses between ML and non-ML approaches. The analysis also explores the utilization of cueing devices. The limited adoption of explainable AI (XAI) approaches in FOG prediction research represents a significant gap. Improving user acceptance and comprehension requires an understanding of the logic underlying algorithm predictions. Current FOG detection and prediction research has a number of limitations, which are identified in the discussion. These include issues with cueing devices, dataset constraints, ethical and privacy concerns, financial and accessibility restrictions, and the requirement for multidisciplinary collaboration. Future research avenues center on refining explainability, expanding and diversifying datasets, adhering to user requirements, and increasing detection and prediction accuracy. The findings contribute to advancing the understanding of FOG and offer valuable guidance for the development of more effective detection and prediction methodologies, ultimately benefiting individuals affected by PD.

A City Shared Bike Dispatch Approach Based on Temporal Graph Convolutional Network and Genetic Algorithm.

Public transportation scheduling aims to optimize the allocation of resources, enhance efficiency, and increase passenger satisfaction, all of which are crucial for building a sustainable urban transportation system. As a complement to public transportation, bike-sharing systems provide users with a solution for the last mile of travel, compensating for the lack of flexibility in public transportation and helping to improve its utilization rate. Due to the characteristics of shared bikes, including peak usage periods in the morning and evening and significant demand fluctuations across different areas, optimizing shared bike dispatch can better meet user needs, reduce vehicle vacancy rates, and increase operating revenue. To address this issue, this article proposes a comprehensive decision-making approach for spatiotemporal demand prediction and bike dispatch optimization. For demand prediction, we design a T-GCN (Temporal Graph Convolutional Network)-based bike demand prediction model. In terms of dispatch optimization, we consider factors such as dispatch capacity, distance restrictions, and dispatch costs, and design an optimization solution based on genetic algorithms. Finally, we validate the approach using shared bike operating data and show that the T-GCN can effectively predict the short-term demand for shared bikes. Meanwhile, the optimization model based on genetic algorithms provides a complete dispatch solution, verifying the model's effectiveness. The shared bike dispatch approach proposed in this paper combines demand prediction with resource scheduling. This scheme can also be extended to other transportation scheduling problems with uncertain demand, such as store replenishment delivery and intercity inventory dispatch.

Deep learning-based approach for 3D bone segmentation and prediction of missing tooth region for dental implant planning

Recent studies have shown that dental implants have high long-term survival rates, indicating their effectiveness compared to other treatments. However, there is still a concern regarding treatment failure. Deep learning methods, specifically U-Net models, have been effectively applied to analyze medical and dental images. This study aims to utilize U-Net models to segment bone in regions where teeth are missing in cone-beam computerized tomography (CBCT) scans and predict the positions of implants. The proposed models were applied to a CBCT dataset of Taibah University Dental Hospital (TUDH) patients between 2018 and 2023. They were evaluated using different performance metrics and validated by a domain expert. The experimental results demonstrated outstanding performance in terms of dice, precision, and recall for bone segmentation (0.93, 0.94, and 0.93, respectively) with a low volume error (0.01). The proposed models offer promising automated dental implant planning for dental implantologists.

Multivariate Hydrological Modeling Based on Long Short-Term Memory Networks for Water Level Forecasting

In the Department of Chocó, flooding poses a recurrent and significant challenge due to heavy rainfall and the dense network of rivers characterizing the region. However, the lack of adequate infrastructure to prevent and predict floods exacerbates this situation. The absence of early warning systems, the scarcity of meteorological and hydrological monitoring stations, and deficiencies in urban planning contribute to the vulnerability of communities to these phenomena. It is imperative to invest in flood prediction and prevention infrastructure, including advanced monitoring systems, the development of hydrological prediction models, and the construction of hydraulic infrastructure, to reduce risk and protect vulnerable communities in Chocó. Additionally, raising public awareness of the associated risks and encouraging the adoption of mitigation and preparedness measures throughout the population are essential. This study introduces a novel approach for the multivariate prediction of hydrological variables, specifically focusing on water level forecasts for two hydrological stations along the Atrato River in Colombia. The model, utilizing a specialized type of recurrent neural network (RNN) called the long short-term memory (LSTM) network, integrates data from hydrological variables, such as the flow, precipitation, and level. With a model architecture featuring four inputs and two outputs, where flow and precipitation serve as inputs and the level serves as the output for each station, the LSTM model is adept at capturing the complex dynamics and cross-correlations among these variables. Validation involves comparing the LSTM model’s performance with linear and nonlinear Autoregressive with Exogenous Input (NARX) models, considering factors such as the estimation error and computational time. Furthermore, this study explores different scenarios for water level prediction, aiming to utilize the proposed approach as an effective flood early warning system.

Game provenance graph-based representation learning vs metrics-based machine learning: An empirical comparison on predictive game analytics tasks

Predicting events and behavior in games is a crucial task of game analytics, both during gameplay, such as predicting wins, and after, like forecasting player churn. Game metrics have been widely adopted due to their human-readable and task-oriented nature. However, these metrics vary between games and require expert knowledge. Game analytics has evolved with techniques that primarily handle non-tabular and sequential data, like graphs and time series. However, existing methods still depend on traditional game metrics from extensive feature engineering. In parallel, graph representation learning has been applied to game provenance graphs — structures that log game events as nodes and link them through causal relationships. This approach may allow analysts to bypass traditional game metrics and learn relevant downstream features during training. This paper compares this metric-less graph-based representation learning with traditional metric-based machine learning in two predictive game analytics tasks: win prediction and hit prediction. Our results show that the same graph-based representation learning architecture achieves competitive results in win prediction and outperforms metric-based approaches in hit prediction, suggesting its suitability for predictive game analytics tasks. This method could significantly benefit as it does not require game-specific metric expertise, and its learned features are versatile across multiple tasks.

Establishing correlates of maternal-fetal cytomegalovirus transmission-one step closer through predictive modeling.

Determinants of maternal-fetal cytomegalovirus (CMV) transmission and factors influencing the severity of congenital CMV (cCMV) infection are not well understood. We conducted a descriptive, multi-center study in pregnant women ≥18 years old with primary CMV infection and their newborns (NCT01251744) to explore maternal immune responses to CMV and determine potential immunologic/virologic correlates of cCMV following primary infection during pregnancy. We developed alternative approaches looking into univariate/multivariate factors associated with cCMV, including a participant clustering/stratification approach and an interpretable predictive model-based approach using trained decision trees for risk prediction (post-hoc analyses). Pregnant women were grouped in three distinct clusters with similar baseline characteristics, particularly gestational age at diagnosis. We observed a trend for higher viral loads in urine and saliva samples from mothers of infants with cCMV versus without cCMV. When using a trained predictive-model approach that accounts for interaction effects between variables, anti-pentamer IgG antibody concentration and viral load in saliva were identified as biomarkers jointly associated with the risk of maternal-fetal CMV transmission. We identified biomarkers of CMV maternal-fetal transmission. After validation in larger studies, our findings will guide the management of primary infection during pregnancy and the development of vaccines against cCMV.

Prediction of Precipitation using Multiscale Geographically Weighted Regression

Prediction of precipitation at locations which lack meteorological measurements is a challenging task in hydrological applications. In this study we aimed to demonstrate potential use of multiscale geographically weighted regression (MGWR) method used to predict precipitation based on relevant meteorological parameters. Geographically weighted regression (GWR) is a regression technique proposed to explore spatial non-stationary relationships. Compared to the linear regression technique, GWR considers the dynamics of local behaviour and, therefore provides an improved representation of spatial variations in relationships. Multiscale geographically weighted regression (MGWR) is a modified version of GWR that examines multiscale processes by providing a scalable and flexible framework. In this study, the MGWR model was used to predict precipitation, which is an essential problem not only in meteorology and climatology, but also in many other disciplines, such as geography and ecology. A meteorological dataset including elevation, precipitation, air temperature, air pressure, relative humidity, and cloud cover data belonging to Türkiye was used, and the performance of the MGWR was assessed in comparison with that of global regression and classical GWR. Experimental evaluations demonstrated that the MGWR model outperformed other approaches in precipitation prediction.

Loss-Based Variational Bayes Prediction

We propose a new approach to Bayesian prediction that caters for models with a large number of parameters and is robust to model misspecification. Given a class of high-dimensional (but parametric) predictive models, this new approach constructs a posterior predictive using a variational approximation to a generalized posterior that is directly focused on predictive accuracy. The theoretical behavior of the new prediction approach is analyzed and a form of optimality demonstrated. Applications to both simulated and empirical data using high-dimensional Bayesian neural network and autoregressive mixture models demonstrate that the approach provides more accurate results than various alternatives, including misspecified likelihood-based predictions. Supplementary materials for this article are available online.

Neuro-Fuzzy Framework for Fault Prediction in Electrical Machines via Vibration Analysis

The advent of Industry 4.0 has ushered in a new era of technological advancements, particularly in integrating information technology with physical devices. This convergence has given rise to smart devices and the Internet of Things (IoT), revolutionizing industrial processes. However, despite the push towards predictive maintenance, there still is a significant gap in fault prediction algorithms for electrical machines. This paper proposes a signal spectrum-based machine learning approach for fault prediction, specifically focusing on bearing faults. This study compares the effectiveness of traditional neural network algorithms with a novel approach integrating fuzzy logic. Through extensive experimentation and analysis of vibration spectra from various mechanical faults in bearings, it is demonstrated that the fuzzy-neuro network model outperforms traditional neural networks, achieving a validation accuracy of 99.40% compared to 94.34%. Incorporating fuzzy logic within the neural network framework offers advantages in handling complex fault combinations, showing promise for applications requiring higher accuracy in fault detection. While initial results are encouraging, further validation with more complex fault scenarios and additional fuzzy layers is recommended to fully explore the potential of fuzzy-neuro networks in fault prediction for electrical machines.

A dynamic hybrid model for efficient tuberculosis incidence rate prediction

ABSTRACT Tuberculosis remains a global health challenge, predicting its incidences is crucial for effective planning and intervention strategies. This study combines AutoRegressive Integrated Moving Average (ARIMA) and Nonlinear AutoRegressive with exogenous input (NARX) models as an innovative approach for TB incidence rate prediction. The performance of the proposed model (ARIMA-NARX) was evaluated using standard metrics (MSE, RMSE, MAE, and MAPE), and it was refined to achieve the best average predictive accuracies with an MSE: 0.0622, RMSE: 0.0851, MAE: 0.07520, and MAPE: 0.05535 followed by NARX 0.1597, 0.3189, 0.2724, and 0.3366, and ARIMA (2,0,0) 0.7781, 0.5959, 0.6524, and 0.6080 Models. These findings are expected to shed light on the TB incidence rate, providing valuable information to policymakers such as the World Health Organization (WHO) and health professionals. The developed model can potentially serve as a predictive tool for proactive TB control and intervention strategies in the region and the world at large.

Synergizing Chemical Structures and Bioassay Descriptions for Enhanced Molecular Property Prediction in Drug Discovery.

The precise prediction of molecular properties can greatly accelerate the development of new drugs. However, in silico molecular property prediction approaches have been limited so far to assays for which large amounts of data are available. In this study, we develop a new computational approach leveraging both the textual description of the assay of interest and the chemical structure of target compounds. By combining these two sources of information via self-supervised learning, our tool can provide accurate predictions for assays where no measurements are available. Remarkably, our approach achieves state-of-the-art performance on the FS-Mol benchmark for zero-shot prediction, outperforming a wide variety of deep learning approaches. Additionally, we demonstrate how our tool can be used for tailoring screening libraries for the assay of interest, showing promising performance in a retrospective case study on a high-throughput screening campaign. By accelerating the early identification of active molecules in drug discovery and development, this method has the potential to streamline the identification of novel therapeutics.

Machine Learning in Assessing Canine Bone Fracture Risk: A Retrospective and Predictive Approach

In the ever-evolving world of veterinary care, the occurrence of bone fractures in canines poses a common and complex problem, especially in extra-small breeds and dogs that are less than 1 year old. The objective of this research is to fill a gap in predicting the risk of canine bone fractures. A machine learning method using a random forest classifier was constructed. The algorithm was trained on a dataset consisting of 2261 cases that included several factors, such as canine age, gender, breed, and weight. The performance of the algorithm was assessed by examining its capacity to forecast the probability of fractures occurring. The findings of our study indicate that the tool has the capability to provide dependable predictions of fracture risk, consistent with our extensive dataset on fractures in canines. However, these results should be considered preliminary due to the limited sample size. This discovery is a crucial tool for veterinary practitioners, allowing them to take preventive measures to manage and prevent fractures. In conclusion, the implementation of this prediction tool has the potential to significantly transform the quality of care in the field of veterinary medicine by enabling the detection of patients at high risk, hence enabling the implementation of timely and customized preventive measures.

An integrated approach for prediction of magnitude using deep learning techniques

An integrated approach for prediction of magnitude using deep learning techniques

The Use of Indigenous Knowledge Systems Practices to Enhance Food Security in Vhembe District, South Africa

AbstractThis paper seeks to examine how subsistence farmers employ indigenous knowledge methods to enhance food security within rural communities in the Vhembe district. It analyses indigenous knowledge practices used in managing climate change impacts on crop production and determines indigenous knowledge processes used to preserve food to fill the gap during food shortages. The paper also considers the indigenous weather prediction approaches used in the area. A participatory rural appraisal is complemented by a qualitative and quantitative approach. This study used semi-structured interviews, questionnaires and focus group discussions for data collection. A survey of 200 randomly selected indigenous farmers was sampled. The finding revealed that local farmers depend on the use of indigenous knowledge practices to improve household food security. The majority of farmers apply manure in their farms using livestock dumps. Also, it shows that 87.8% of households used sun-drying processes for food preservation and used local mortars to pound and grind foodstuffs such as peanuts and maize. Practices such as crop rotations, mixed cropping and intercropping were employed to improve soil fertility and climate change and reduce insect pest outbursts on crops, hence, improving crop production. Indigenous knowledge of rainfall prediction is helpful in preparation for the planting season. The study recommended that agriculture policies must acknowledge indigenous knowledge practices in development programmes and specific policy interventions to promote the indigenous knowledge systems must focus on enhancing socio-economic factors assisting farmers in improving post-harvest storage facilities. The findings could facilitate interaction between indigenous knowledge stakeholders and the Limpopo provincial agriculture training institution. The fulfilment of the United Nations Sustainable Development Goal 2 on zero hunger and food security could benefit from the indigenous knowledge systems.