Due to unchecked cell development in the lungs, lung cancer typically affects both men and women. This seriously impairs one's ability to breathe in and out of the chest. According to the World Health Organisation, the main causes of lung cancer are cigarettes and passive smoking. Compared to other cancers, the death rate from lung cancer is rising daily among both young and old people. Despite the availability of advanced medical facilities for accurate diagnosis and effective medical care, the mortality rate is still not effectively under control. The prediction of lung cancer was investigated using classification techniques as Naive Bayes, Random Forest, Decision trees, and Logistic Regression. The primary objective of the research is to analyse the effectiveness of categorisation algorithms in order to identify lung cancer early.

Predicting stock market trends is crucial for investors and policymakers, particularly in emerging markets like the Middle East and North Africa (MENA) region, where research on stock market prediction remains relatively scarce. This study provides a comparative analysis of several advanced forecasting models, including Autoregressive Integrated Moving Average with Explanatory Variable (ARIMAX), Support Vector Regression (SVR), Extreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), and HistGradientBoosting (HGB), to evaluate their effectiveness in predicting stock market movements in the region. We utilized the Skforecast library to implement the ARIMAX model and applied the Grid Search algorithm for hyperparameter tuning of SVR and the XGBoost family of models. Model performance was assessed using a comprehensive dataset and a suite of metrics: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Normalized Mean Square Error (NMSE), R-squared ( R^2 ), Theil’s U-statistic, Directional Symmetry (DS), and Weighted Directional Symmetry (WDS). Additionally, we analyzed the interpretability of the XGBoost model using Shapley Additive exPlanations (SHAP) and evaluated overall model performance with a Taylor diagram. Our findings indicate that both ARIMAX and SVR outperform the other models in terms of predictive accuracy and reliability. However, SVR emerges as the best model, achieving the lowest MAE and RMSE, the highest R^2 , and gives the most favorable results for the NMSE, DS, WDS, and Theil’s U statistic. This study highlights the superior performance of SVR for forecasting stock market trends in the MENA region, providing valuable insights for financial analysts and investors.

Traffic congestion forecasting is one of the major elements of the Intelligent Transportation Systems (ITS). Traffic congestion in urban road networks significantly influences sustainability by increasing air pollution levels. Efficient congestion management enables drivers to bypass heavily trafficked areas and reducing pollutant emissions. However, properly forecasting congestion spread remains challenging due to complex, dynamic, and non-linear nature of traffic patterns. The advent of Internet of Things (IoT) devices has introduced valuable datasets that can support the development of intelligent and sustainable transportation for modern cities. This work presents a Deep Learning (DL) approach of Reinforcement Learning (RL) based Bidirectional Long Short-Term Memory (BiLSTM) with Adaptive Secretary Bird Optimizer (ASBO) for traffic congestion prediction. The experimentation is evaluated on Traffic Prediction Dataset and achieved better Mean Square Error (MSE) and Mean Absolute Error (MAE) with results of 0.015 and 0.133 respectively. Compared to the existing algorithms like RL, Deep Q Learning (DQL), LSTM and BiLSTM, the RL - BiLSTM with ASBO outperformed with the parameters MSE, RMSE, R2, MAE and MAPE with 37%, 27.44%, 26%, 33.52% and 35.8% respectively. The better performance demonstrates that RL- BiLSTM with ASBO is well-suited to predict congestion patterns in road networks.

Most critically, enforcing road safety laws and reducing roadway fatalities from traffic accidents will rely heavily on accurately detecting helmet use and motorcycle riders' number plates. Conventional methods of enforcement rely heavily on human observation and labour-intensive methods, and are problematic as they can be subject to human error under specific conditions (for example, poor weather and lighting). To demonstrate an automated system that could potentially mitigate some of these constraints, we developed a novel automated system by combining cutting-edge object detection models (You Only Look Once, Nano version (YOLOv8n), Faster R-CNN) and EasyOCR (optical character recognition) to assist law enforcement. YOLOv8n has been praised for its speed and accuracy; when developing and training the model, we sourced a dataset with images of riders wearing helmets and other images of riders without helmets, as well as the number plate images to ensure that during training, the model was able to learn both tasks as it related to the helmet use detection and number plates. EasyOCR is an excellent resource for extracting textual data from detected number plates, so combining YOLOv8n with EasyOCR allows for real-time data collection from the detected number plates. This unified platform is a big step ahead in roadway security technology. It offers an easily expandable yet successful way to keep an eye on helmet use and identify and punish drivers who don't wear them, all of which will help reduce traffic accidents and save lives.

Considering that all moving objects are humans, much of the work in data is based on recognizing and tracking moving objects. In this work, we present a method for counting people's faces. Even though we use the face mask, the deep learning-based YOLOv5 algorithm and Faster R-CNN allow us to recognize the face. We do a very good job of counting people. To make the calculation more accurate, we introduced a new type of intelligent small scale computing system consisting of cheaper hardware and user-friendly cloud computing software. These findings show that intelligent computing systems can realize human vision. Additionally, by combining inexpensive hardware with cloud-based software, the planning process becomes more transparent and cost-effective. Finally, the web application allows users to view the number of authorized and unauthorized users. Based on the results obtained from this method, the deep learning YOLOv5 algorithm is used to identify and match human images to increase security, and thanks to cloud storage, users can easily view all calculated results, increasing the accuracy by 98.53

The rapid expansion of digital media and the seamless transmission of information have raised serious concerns about the widespread dissemination of misinformation and fake news. Combatting this issue requires robust and effective techniques that can accurately detect and classify fake news. Natural language processing (NLP) approaches have emerged as powerful tools in this endeavor, leveraging advanced text classification algorithms to identify and counteract misinformation. This study includes NLP approaches for countering misinformation through text classification, with a specific focus on fake news detection. Leveraging natural language processing techniques, the project implements a text classification pipeline for identifying and distinguishing between genuine and fake news. The pipeline encompasses essential NLP steps such as tokenization and stop word removal. Traditional machine learning algorithms, such as the gradient boosting classifier, CatBoost classifier, random forest classifier, AdaBoost classifier, logistic regression, and SVM linear kernel are trained using the transformed data to classify news articles. This study explores feature engineering techniques and model evaluation to enhance the classification performance. Experimental results indicate the effectiveness of this NLP-based text classification approach in detecting fake news, demonstrating promising accuracy rates out of which gradient boost classifier gives the highest accuracy rate. The simplicity and accessibility of the proposed method make it an ideal starting point for beginners in NLP, offering insights into the application of NLP techniques for countering misinformation and promoting information integrity.

Identifying brain abnormalities via Magnetic Resonance Imaging (MRI) scans is pivotal when considering such conditions as multiple sclerosis (MS) and gliomas. Medical imaging segmentation techniques are often reliant on large volumes of annotated data, which is usually expensive and time consuming to accumulate. This study proposed an unsupervised detection for both anomaly detection and segmentation of brain MRI segmentations primarily using autoencoders (AEs), variational autoencoders (VAEs) or Generative Adversarial Networks (GANs), while at the same time reducing the need for labelled datasets. In this context, these machines are trained on MRI scans of healthy brains, to understand the anatomical structures and patterns that are considered normal, so that the models can identify an abnormality by observing an anomalous behaviour or change to what was previously thought to be normal. The methodology used these models to segment brain MRI scans to identify anomalies or differences in the abnormal brain MRI scans compared to their normal structure from the training set. A comparative study was undertaken across three MRI datasets with images from healthy patients, MS lesions and glioma. The three deep learning models of AE, VAE and f-AnoGAN (a GAN approach to anomaly detection) were evaluated. The experimental results indicated that the VAE recorded a segmentation accuracy (of import) of 91.8% for all MRI images analysed, while the f-AnoGAN model recorded an accuracy of 85.2% when identifying an abnormality in MRI images scans. Additionally, both the VAE and f-AnoGAN models outperformed the traditional autoencoder, which failed to detect the anamoly. As such, the models offer viable options for aiding clinicians diagnose neurological diseases as well as simplifying MRI segmentation procedures.

Improving urban cities' quality of life is one of the primary functions of urban green spaces. Efficient lighting solutions in these areas enhance safety and ambiance. This research introduces a new method for autonomously controlling the lighting in urban parks with the help of a Naive Bayes classifier and cloud computing. The system uses information gathered from several sensors to adapt the brightness of the lights in response to changes in the surrounding environment and the presence or absence of people. The system can generate accurate choices in real-time using machine learning methods, including the Naive Bayes algorithm. With the help of the cloud, remote monitoring and administration are made easier, which improves efficiency and scalability. The experimental findings prove that the proposed approach is reliable and efficient in enhancing urban green areas' energy efficiency and user enjoyment. Intelligent algorithms combined with cloud computing provide a viable option for smart city lighting control, opening the door to more responsive and environmentally friendly cityscapes.

This research focuses on sustainable crop protection through the integration of the Internet of Things (IoT) and drone-based spraying. The main goal is to improve agricultural practices' accuracy, efficiency, and environmental friendliness. Drones with spraying and IoT capabilities allow for the precise and controlled management of crop protection chemicals like insecticides. This approach minimizes the use of chemicals, reduces their environmental impact, and makes the most efficient use of resources. When used with IoT sensor data, support Vector Machine (SVM) analysis improves decision-making by providing predictive insights. SVM analysis allows for the accurate localization of treatment zones, which improves pesticide application while reducing damage to non-target species. This work proposes a potential way forward for long-term crop security using SVM analysis and IoT-enabled drone technology. It signals a new era of environmentally conscious crop protection techniques prioritizing effective pest control without sacrificing environmental preservation.

This research study proposes a novel approach to enhance hospital lighting management by implementing dynamic lighting control. Traditional lighting systems often fail to meet the specific needs of patients, potentially impacting their health and well-being. The proposed system integrates the Internet of Things (IoT) and Reinforcement Learning (RL) algorithms to continuously track environmental factors and patient input. By analyzing this data, the system can make real-time adjustments to lighting characteristics, including intensity, color temperature, and spatial distribution. The proposed system’s autonomous optimization of lighting settings aims to promote patient comfort, mood, and overall well-being. Simulation results demonstrate the effectiveness and practicality of this approach in revolutionizing hospital lighting control. By prioritizing patients and focusing on individual needs, this research contributes to the development of healthcare facilities that enhance patient experience and promote healing.

As urban areas like Chennai and Bangalore witness a continuous surge in land and housing prices, accurately estimating the market value of houses has become increasingly crucial. This presents a formidable challenge, prompting a growing demand for an accessible and efficient method to predict house rental prices, ensuring dependable forecasts for future generations. In response to this need, this study delves into the core factors influencing rental prices, with a keen focus on location and area. Leveraging a dataset comprising ten essential features tailored for detecting Rental Price in Metropolitan cities, the research meticulously preprocesses the data using a Python library to ensure data cleanliness, laying a robust foundation for constructing the predictive model. Employing a diverse range of Machine Learning algorithms, including Random Forest, Linear Regression, Decision Tree Regression, and Gradient Boosting, the study evaluates their efficacy in forecasting rental prices. Notably, feature extraction underscores the significance of area and property type in shaping rental prices. In comparison with existing methodologies, this research adopts gradient boosting as its preferred approach, achieving the most satisfactory predictive outcomes. Evaluation metrics are meticulously analyzed to validate the model's performance. Through this comprehensive analysis, the study not only offers valuable insights into rental price prediction but also ensures a rigorous comparison with existing approaches, maintaining originality and relevance in addressing the pressing challenges of housing market dynamics.

Alzheimer’s Disease (AD) is a degenerative, chronic condition of the brain for which there is now no effective treatment. However, there are medications that can slow its development. In order to stop and control the development of AD, earlier diagnosis of the disease is quintessential. Our proposed method’s primary objective is to establish a comprehensive model for the prior detection of Alzheimer’s disease and the categorization of distinct AD stages. This work employs a deep learning methodology, especially CNN. The proposed approach makes use of well-known models that have already been trained to classify medical images, like the EfficentNetB7 model, by applying the transfer learning principle. In order to achieve greater accuracy, convolutional neural networks (CNNs) are frequently scaled up as new resources become available at a fixed cost throughout the construction phase. A compound coefficient is used by the CNN architecture and scaling approach, which is the foundation of the pre-trained EfficientNetB7 model, to scale the dimensions equally. This proposed EfficientNetB7 model is quicker, easier, and more effective than other pre-trained models like VGG19 and InceptionV3. The proposed model includes simple structures that have memory requirements, provide manageable time, overfitting, and low computational complexity as well as training and inference speeds. The Alzheimer’s disease Neuroimaging Initiative (ADNI) dataset was employed for a comprehensive assessment of the method proposed, utilizing well-known performance metrics including sensitivity, specificity, and accuracy. The findings revealed that the improvised results achieved the accuracy metric when compared to existing methods. The EfficientNetB7 model has been enhanced, and this model achieves a sensitivity of 98.08

Traditional surveillance systems in public spaces rely heavily on 24/7 human monitoring, which can be labourintensive, prone to fatigue, and subject to human error, thereby potentially compromising security effectiveness. This work addresses these limitations by proposing an innovative IoT-based solution utilising the YOLO (You Only Look Once) model for real-time video analysis. The primary objective is to automate and enhance surveillance capabilities through the integration of IoT cameras with edge computing devices equipped with GPU acceleration. This setup enables local processing of video streams, allowing the YOLO model to swiftly and accurately detect various suspicious activities such as loitering, unauthorised access, or other predefined anomalies. Alerts generated by the YOLO model are efficiently communicated via a chatbot interface, facilitating immediate responses from security personnel or automated systems. By reducing the reliance on continuous human monitoring, the system improves operational efficiency and responsiveness to security incidents. It minimises the risk of oversight and delays in detecting critical events, thereby enhancing overall security measures in public premises. The scalability and adaptability of the proposed system make it suitable for deployment in diverse environments, from small-scale installations to large public areas, ensuring comprehensive surveillance coverage and proactive threat detection. This IoT-based approach leveraging advanced video analytics with the YOLO model represents a significant advancement in enhancing security surveillance capabilities. It offers efficient real-time detection, rapid response mechanisms, and scalability, ultimately contributing to safer public spaces and better resource utilisation in security management.

In the current scenario, people are suffering and isolated by themselves by seizure detection and prediction in epilepsy. Also, it is highly essential that it needs to be identified through wearable devices. Researchers discussed this issue and outlined future developments in this field, suggesting that Machine Learning (ML) techniques could radically change how we diagnose and manage patients with epilepsy. However, as data availability has increased, Deep Learning (DL) techniques have become the most cutting-edge approach to adopt and use with wearable devices. On the other hand, large amounts of data are needed to train DL models, making overfitting problematic. DL models are created with open-source toolboxes and Python, allowing researchers to create automated systems and broaden computational accessibility. This work thoroughly overviews deep learning (DL) methods and neuroimaging modalities for automated epileptic seizure identification. It covers several MRI and EEG techniques for epileptic seizure diagnosis and treatment programmes designed to treat these seizures. The study also covers the difficulties in precise detection, the benefits and drawbacks of DL-based strategies, potential DL models and upcoming research in this area.

A potential approach to improve firefighters' performance in difficult emergencies has been the combination of exoskeletons with Internet of Things (IoT) technologies in recent years. To provide adaptive assistance to firefighters during crucial missions, this research presents a new framework that uses reinforcement learning (RL) algorithms combined with IoT exoskeletons. The proposed system uses interconnected sensors built into the exoskeletons to track various environmental and physiological variables, including core temperature, heart rate, and ambient light intensity. Firefighters' health and the danger level of the environment are evaluated using these data points in real time. Autonomously adjusting their support systems to offer appropriate help based on the dynamic situation, the exoskeletons use RL techniques. The system learns to predict the actions of firefighters, adapt the amount of help in real-time, and maximize energy efficiency to keep running for longer due to feedback and iterative learning processes. Firefighters and incident command centers can communicate seamlessly with the proposed framework, allowing real-time situational awareness and decision assistance. Firefighters' ability to be safe and successful during emergency operations is greatly enhanced by exoskeletons. Smart solutions can protect first responders from harm and make them more resilient in dangerous situations.

Air pollution is described as the contains of dangerous substances in atmosphere that are emitted by natural sources or human activities and can harm humans, animals, and the environment. Sulfurous anhydride, carbon monoxide, Dioxide of nitrogen, trioxygen, carbon monoxide, and volatile organic compounds are some of the most common air contaminants. The development of tools for tracking and reducing air pollution is greatly supported by this analysis, which is important for both public health and environmental sustainability. To improve model performance, particular attention is paid to feature engineering and data pre-treatment methods. The study employs machine learning models, including Decision Tree Classifier, Logistic Regression, Naive Bayes, K-Nearest Neighbors, Random Forest, and SVM, to ascertain the most effective approach for accurate air quality forecasting. The key contribution of this work lies in its ability to provide a practical decision-making framework that can guide authorities in implementing focused interventions and optimizing resource allocation for pollution control. The hybrid approach utilized in this study surpasses traditional methods, offering a more precise tool for air quality forecasting. This model not only addresses the need for improved predictive performance but also acts as a useful tool for public health and environmental agencies seeking to minimize the harmful effects of air pollution.

Wound treatment is an essential part of patient care that requires continuous training and education for medical personnel. Using Long Short-Term Memory (LSTM) networks for time-series feedback and Internet of Things (IoT), technologies provide a paradigm shift in wound education in this research. The platform combines LSTM-based learning algorithms with real-time data collected from IoT wound monitoring to provide healthcare providers with ongoing training and feedback. A battery of trials illustrates that strategy outperforms the status quo regarding wound assessment accuracy and treatment decision-making. This approach promotes ongoing professional growth in wound care management by allowing for individualized learning experiences. Its work shows that continuous training using LSTM powered by the IoT has great promise for better wound education and patient outcomes in healthcare.

Artificial Intelligence is deployed in multiple areas, including healthcare. Utmost research is done in AI enabled healthcare industry because of the demands like accurate result, data security, exact prediction, huge volume of data, etc. In conventional deep learning models, the training happens with the dataset that are stored in a single device. This requires a huge storage space and highly efficient machines to train the data. Usage of big data, demands for innovative models that can be deployed and used in confined storage. Split learning is one such collaborative distributed deep learning model that allows the data to be stored in a split fashion. Split learning supports desirable features like less storage, more privacy to raw data, ability to work with resource constraints, etc., making it suitable for storing electronic medical record of patients. This paper discusses the advantages of using split learning for healthcare, the possible configurations of split learning that supports data privacy in healthcare and finally discusses the open research challenges in implementing split learning for healthcare.

Sign language detection, specifically for Indian Sign Language (ISL) and American Sign Language (ASL), plays a crucial role in bridging communication gaps between hearing impaired and normal people..This paper proposes a systematic methodology using different deep learning models to detect ISL and ASL from an image dataset of sign language gestures. The algorithms probed were 2D Convolutional Neural Networks (CNN), ResNet-50, Long Short Term Memory (LSTM) and VGG16. By using a dataset we conducted an extensive comparative analysis of these algorithms. The results portrayed LSTM as the best predictive classifier with an amazing 97.97% accuracy. This excellent result makes LSTM a strong model for the classification of sign languages with high accuracy. The target classes involved several signs from ISL and ASL are thus extensive for recognizing and interpreting both these languages by using a number of algorithms we make the predictive model more robust as it can handle multiple kinds of data patterns corresponding to different sign gestures. The outcomes of our investigation present the very first step towards an intelligent sign language detection system. A well-recognized and fast-emerging potential of deep learning is in the automatic detection of sign languages, which introduces a drastic positive change in communication efficiency as this research field grows exponentially. This could greatly improve the ability of a person to interact and the accessibility of an individual to sign languages.

This research introduces a new method for predicting electric vehicles (EVs) range that combines cloud computing with random forest regression (RFR) approaches. Predicting the range properly is now critical for user comfort and efficient route planning due to the rising popularity of EVs. Factors like changing driving conditions and battery depletion might make traditional range calculation techniques inaccurate. It uses the processing power of the cloud to collect and analyze massive volumes of real-time data on variables like traffic, weather, and topography that have an impact on EVs range. Next, it uses a powerful machine learning (ML) algorithm for RFR to accurately forecast EVs range by modeling the intricate interactions between these factors. Results from experiments this technique works, with far better range estimate advanced approaches. Cloud computing scalability and flexibility also make it easy to predict models into preexisting EVs navigation systems, so drivers may get accurate range estimations in real time based on their own circumstances. It will improve EVs’ usability and dependability, which will lead to more people buying and using these eco-friendly vehicles.