Parkinson's disease (PD) is a degenerative neurological disease, and at present there are no reliable laboratory tests for it. So how does this happen when people go to identify PD? vocal biomarkers, combined with machine learning (ML), seem to be an option for noninvasive diagnostics. In our work, we used a voice recording dataset which consisted of 26 different feature sets mined by various techniques. When using the extreme gradient boosting (XGBoost) method, out of all these models tested, an accuracy of 91.79% was achieved. As can be seen from its high precision, recall and F1- score, XGBoost performed very well in differentiating PD cases from non-cases. The study concludes that the application of ML, particularly XGBoost, to the diagnostic process can establish a valuable tool for early screening of PD, which will facilitate more speedy and correspondingly cost-effective clinical evaluations. This paper represents an important contribution to the rapidly developing fields of artificial intelligence-based on diagnosis of neurological diseases and digital health.

Chronic obstructive pulmonary disease (COPD) is a global health problem, which requires accurate and efficient diagnostic methods to enable timely and effective treatment. In this study, we propose a novel method for COPD detection that uses time-frequency (TF)-based deep neural networks (DNNs) for lung sound (LS) signal classification. We utilized several TF representation (TFR) methods, including continuous wavelet transform (CWT), Wigner-Ville distribution (WVD), and smoothed pseudo WVD (SPWVD) to represent the LS signals in the TF plane. In the next step, the TFR images are used as input to train pretrained convolutional neural networks (CNNs), namely, EfficientNet-B0, ShuffleNet-V2, MobileNet-V2, and GhostNet for COPD detection. We have also performed the multiscale analysis of the proposed framework by employing the variational mode decomposition (VMD) method. The VMD decomposes LS signals into intrinsic mode functions (IMFs) that facilitate multiscale TFR, which serves as input to the DNN models for COPD detection. The proposed method's performance is evaluated on publicly available ICBHI 2017 and Fraiwan datasets utilizing different performance metrics. In the proposed framework, CWT-based TFR with EfficientNet-B0 achieves superior performance over other models, attaining the highest accuracies of 98.57% and 95.10% in detecting COPD on ICBHI 2017 and Fraiwan's datasets, respectively. The interpretability of the proposed framework is demonstrated using t-distributed stochastic neighbor embedding (t-SNE) plots considering all the TFR methods and CNN models. Furthermore, a statistical analysis of the multiscale deep features is performed using the Kruskal-Wallis statistical test. Finally, the model is deployed as a standalone system on an edge device, enabling real-time inference and classification.

This paper proposes an approach to compute low-variance, translation-insensitive features from multivariate electroencephalogram (EEG) signals in different oscillatory levels for schizophrenia (SZ) detection. We explore the wavelet scattering transform (WST), which is structurally reminiscent of convolutional neural networks (CNNs), for sparse representation of multivariate EEG signals. The empirical wavelet transform (EWT) has been employed for decomposing multivariate EEG signals into rhythms (S, B, a, /i, and y) for multi-scale analysis. We use WST to encode the hierarchical structure of EEG rhythms by computing low-variance and shift-invariant features. Finally, the scattering features are fed to a long short-term memory (LSTM)-based recurrent deep neural network (DNN) for classification of healthy control (HC) and SZ EEG signals. The model effectively discriminates between SZ and HC EEG signals, achieving the highest classification accuracies of 99.127%, 98.81%, and 93.75% on IBIB PAN, Mental Health Research Center (MHRC), and Basic Sensory Task EEG datasets, respectively. Extensive experiments were carried out to rigorously assess the statistical significance of WST, as well as the robustness and generalizability of the proposed method. The proposed method demonstrates state-of-the-art performance for SZ detection and has been deployed as a standalone MATLAB desktop application.

In this study, we propose a deep learning (DL)-based system for the real-time classification of phonocardiogram (PCG) signals, implemented on an embedded Raspberry Pi platform. The system leverages a publicly available PCG dataset of heart sound recordings, which are converted into time-frequency representations (TFRs) using the continuous wavelet transform (CWT). These TFRs serve as inputs to finetuned, pre-trained convolutional neural network (CNN) models, including VGG16, ResNet50, MobileNetV2, and DenseNet121. Among these, ResNet50 achieved the highest classification accuracy (91.38%) and F1-score (86.95%), while MobileNetV2 demonstrated the fastest inference time on the Raspberry Pi, emphasizing its suitability for deployment in resource-constrained environments. Experimental results confirm that the proposed system delivers both accurate and near-real-time classification of PCG signals, highlighting its potential as an effective and affordable solution for cardiovascular screening, especially in rural and remote areas.

The incorporation of machine learning techniques in medical research has facilitated the exploration of novel avenues for the timely identification of diseases. The continuous progress in medical technology has facilitated the acquisition of complex and complete datasets, which in turn enhances the ability to identify medical diseases in their early stages. Alzheimer’s disease, a significant and hard problem, is characterised by the slow degeneration of brain cells and has a profound impact on cognitive functions, namely memory. It occupies a prominent position within this domain. In the middle of these exciting promises, there remains a significant research gap that pertains to the absence of thorough empirical evidence about the effectiveness of machine learning algorithms in the early identification of Alzheimer’s disease. The primary objective of this study is to address the existing research gap by conducting a comprehensive and meticulous series of experiments. A comprehensive examination of data obtained from sophisticated neuroimaging technologies is performed by utilising a wide range of machine learning models, such as Logistic Regression, Naive Bayes, Neural Networks, Random Forest, and the Stack ensemble. The primary objective is to facilitate the prompt detection of Alzheimer’s disease, hence enabling expedited interventions and therapeutic approaches. As one embarks on the journey of research, the unfolding narrative is shaped by the use of empirical evidence, establishing a strong foundation in the convergence of state-of-the-art technology and the urgent healthcare need to detect early stages of Alzheimer’s disease. Furthermore, this research not only addresses existing gaps in the literature but also ends in the identification of the most effective machine learning model, specifically the Neural Network, which has an accuracy rate of 87

This research endeavours to predict stress levels in sleep patterns through the innovative utilization of a Smart Yoga Pillow (SaYoPillow) combined with machine learning models. In today’s fast-paced lifestyle, stress management is paramount for overall well-being. However, people often overlook the significance of sleep in mitigating stress. The SaYoPillow is an innovative gadget that gathers many physiological measurements while a person is sleeping, such as the frequency of snoring, the rate of breathing, body temperature, the speed of limb movement, blood oxygen levels, eye movement, duration of sleep, and heart rate. Leveraging this data, our objective is to predict stress levels for the subsequent day, thereby materializing the concept of “Smart-Sleeping”. The research addresses the gap in existing stress management methodologies by offering a proactive and personalized approach rooted in real-time physiological monitoring. Our findings demonstrate the efficacy of various machine learning models, with classifiers such as KNN, Logistic Regression, and Gaussian Naive Bayes achieving 100% accuracy in stress level prediction. The results underscore the potential of the SaYoPillow system in revolutionizing stress management practices, enabling individuals to make informed decisions to enhance their sleep quality and overall well-being.

Early detection of rice plant diseases could help to quickly eradicate numerous diseases, such as fungi, viruses, and bacteria, consequently increasing rice yield. Traditional techniques for performing this task may not be the best because they take a long time, require experienced personnel, and are susceptible to a variety of infections. As a result, machine learning and deep learning approaches have recently been utilized to overcome these issues and present a more accurate model for detecting rice plant diseases. However, the current machine learning (ML) and deep learning (DL) models for this task produce unsatisfactory results due to many constraints, including high computational expenses and overfitting. To address these limitations and obtain more accurate disease detection for rice plants, we present a hybrid model of MobileNet and DNN (HMobileNetDNN). The small size of MobileNet minimizes computing costs, while the depth and complexity of DNN improve the model's capacity to capture complicated features, yielding satisfying results. Furthermore, the proposed HMobileNetDNN is also compared to four transfer learning-based DL models, namely ResNetV2, InceptionV3, MobilenetV2, and DensNet121, using the Paddy Doctor dataset. We employ several performance metrics to assess the effectiveness and efficiency of the models, like accuracy, precision, recall, F1 score, and area under the curve. The proposed model outperformed the comparing models, achieving values of 0.918, 0.918, 0.907, 0.912, and 0.949 for accuracy, precision, recall, F1 score, and area under the curve, respectively.

Bharat is a cultivation-based country that cultivates various crops, including rice, in colossal capacity. Rice is one of the daily used crops across the country and other parts of the globe. Rice is cultivated in almost every state in Bharat. However, rice plant diseases severely degrade the quality and quantity of the crop. Rice plants can be affected by different conditions, e.g., foot rot, sheath blight, etc., causing a loss in the farming output. Various steps, including academic research, are practiced to detect rice diseases in their early stage to minimize the loss and increase the productivity of the agriculture sector. The schemes involve traditional and technology-based innovative solutions like agricultural cyber-physical systems (ACPS) or precision agriculture. This paper provides a state-of-the-art on rice crop cultivation, its disease classifications, etc. The paper depicts the importance of rice in daily life and how it plays a vital role in health. Later, this paper provides statistics on rice production in Bharat, followed by the export of rice to different countries. Subsequently, we offer a detailed analysis of rice cultivation, various diseases that can occur in rice plants and other prevention mechanisms for these diseases in the early stages of their life cycle. Finally, recent developments in precision agriculture are provided.

Rice is a staple food in Bharat (India) and many other parts of the world. However, the increasing demand for rice due to population growth forces various challenges, including degraded crop quality and quantity due to rice plant diseases. Diseases such as brown spots, bacterial blight, and hispa can significantly reduce farming output, thereby impacting the productivity of the agriculture sector. To address this challenge, various solutions such as Agricultural cyber-physical systems (ACPS) and precision agriculture have been proposed, along with the application of deep learning techniques. This paper presents a rice leaf disease detection method using deep transfer learning. The proposed approach explores well-known pre-trained deep Convolutional Neural Network (CNN) models - VGG19, DenseNet201, InceptionV3, ResNet50, EfficientNetB3, EfficientNetB7, and XceptionNet, for image-based rice disease classification. Experimental results show that the DenseNet model by the proposed method achieved the highest classification accuracy of 98.75% when fine-tuned properly. The proposed scheme outperforms many existing approaches, delivering a superior disease control solution for rice leaf diseases.