Skin cancer (SC) is the most prevalent cancer globally. Clinical assessment of skin lesions is needed to assess the disease characteristics; but, it is varied in interpretation and limited by longer timelines. Dermoscopy refers to a non-invasive imaging method that permits dermatologists to examine skin lesions with improved visualization of surface and subsurface structures. Machine-learning (ML) and deep learning (DL) methods were established to support dermatologists and overcome the problem as accurate and earlier diagnoses of SC is critical to increasing survival rate of the patient. In recent years, SC classification on dermoscopy images using DL has received great deal of interest. DL algorithms, especially convolutional neural networks (CNNs), have proved promising outcomes in precisely classifying skin lesions and differentiating between benign and malignant cases. The study presents a new Dung Beetle Optimization Algorithm with Multi-modal Deep Learning based Skin Cancer Classification (DBOA-MMDLSCC) technique on dermoscopic images. The presented DBOA-MMDLSCC technique exploits the ensemble learning mechanism by the use of three DL models with metaheuristic based hyperparameter tuning. In the presented DBOA-MMDLSCC technique, U-Net++ model is utilized for skin lesion segmentation. For feature extraction, multi-modal DL model comprising three DL models namely Xception, Residual Network (ResNet), and SqueezeNet. Moreover, the hyperparameter tuning of the DL approaches takes place using the DBOA. Lastly, convolutional autoencoder (CAE) model is applied for detecting and classifying SC. A wide range of experiments were performed to exhibit the simulation results of the DBOA-MMDLSCC technique. The experimental values highlighted the improved performance of the DBOA-MMDLSCC technique in terms of different evaluation measures.

Pedestrian detection using object detection and deep learning has been found to be effective method for identifying pedestrians in video frames or images accurately. It is more commonly used in many real-time applications, such as security observing systems, autonomous driving systems, and robotics. The combination of deep learning techniques and object detection algorithms allows efficient and robust detection of pedestrians in several real-time scenarios. But it is necessary to improve the detection efficacy for complex environments such as cases with worse visibility due to weather or daytime, crowd scenes, and rare pose samples. Continuous improvement and research in DL algorithms, dataset collection, and training models contribute to accelerating the robustness and accuracy of pedestrian detection systems. Therefore, this study designs a new marine predator algorithm with deep learning-based pedestrian detection and classification (MPADLB-PDC) technique. The objective of the MPADLB-PDC approach lies in the accurate recognition and classification of pedestrians. To achieve this, the MPADLB-PDC technique involves two major processes, namely object detection and classification. In the first stage, the MPADLBPDC technique uses an improved YOLOv7 object detector for the recognition of the objects in the frame. Next, in the second stage, the ensemble classifier comprises three classifiers such as deep feed-forward neural network (DFFNN), extreme learning machine (ELM), and long short-term memory (LSTM). To improve the recognition performance of the ensemble classifier, the MPA is used to optimally select the parameters related to it. The simulation outcome of the MPADLB-PDC system was validated on the pedestrian database, and the outcome can be studied interms of various aspects. The simulation values validated the better outcome of the MPADLB-PDC system compared to other approaches.

An immediate requirement in modern medicine is for computer-aided detection and diagnosis of bone fractures. Radiologists can save time and work more efficiently as a result. In the past, numerous different image processing methods were utilized to spot bone breaks. In the field of medical image processing, deep learning models in the form of specialized convolutional neural networks are currently in widespread usage. Its scope is broadened to include bone fracture detection in X-Ray images. The severity of injuries can be estimated by extracting information such as the presence of fractures, their locations, and the distances in length, width, and depth between the broken bones in an automated method. We present methods for identifying bone fractures in diagnostic imaging. The major objective is to use multiple deep learning models to detect these cracks. Patients' elbow, hand, and foot X-rays are separated into their own categories in the dataset. The proposed system analyzes or diagnoses the output using frame difference and a data analytics approach. All models have their feature extraction handled by the proposed method. Because it double-checks feature extractions, it rarely fails to detect a fracture. The proposed system is worn on the hand's wrist. This software tells users the state of their fractures once they answer a few simple questions and upload an X-ray. The image is subsequently processed by the application's built-in models, with all results (whether or not a fracture was identified) shown in the user interface. Python scripts are incorporated into the application's C#-based Dotnet framework. Better sensitivity and specificity are achieved, with an overall accuracy of 89%. Correct diagnosis is crucial since wrong diagnoses can have devastating effects on patients.

Today, sepsis affects several individuals in the Intensive Care Unit (ICU)because the death rate is increasing dramatically and it has become a huge concern in the area of healthcare. Due to the scarcity of resources, such persons require significant upkeep; this increases the expense of therapy by consuming a large number of resources. In early stages of sepsis, therapy is accessible; however, if treatment is not initiated at the appropriate time, sepsis progresses to an advanced stage, increasing deaths. Several investigations are conducted in order to establish early detection models for sepsis in patients. Machine Learning (ML) and Artificial Intelligence (AI) comprise several applications in the medical area, thanks to breakthroughs in these domains. In this paper, five recent ML approaches, including Decision Tree (DT),Support Vector Machine (SVM), Logistic Regression (LR),Nave Bayes (NB)and-Nearest Neighbour (KNN)is engaged to improve the prediction of model performance utilizing a suggested Stacking Ensemble Meta (SEM) algorithm. Then the suggested SEM combines the best of the two base model prediction accuracy to produce the final prediction. The model is trained using data from the 2019 Physio Net/ Computing in Cardiology Challenge. The model is fine-tuned to get the optimal hyper parameters for training. Accuracy, Recall, Precision, F1-score and ROC-AUC curve are few performance measures utilized to assess the model. The experimental findings showed that the suggested SEM classifier has 0.94 accuracy when compared to the ensemble voting classifier, which has 0.93 accuracy.

No abstract