Content uploaded by Mohsen Ghaffari
Author content
All content in this area was uploaded by Mohsen Ghaffari on Oct 18, 2022
Content may be subject to copyright.
A preview of the PDF is not available
Learning-based systems for assessing hazard places of contagious diseases and diagnosing patient possibility
Abstract
To manage the propagation of infectious diseases, particularly fast-spreading pandemics, it is necessary to provide information about possible infected places and individuals, however, it needs diagnostic tests and is time-consuming and expensive. To smooth these issues, and motivated by the current Coronavirus disease (COVID-19) pandemic, in this paper, we propose a learning-based system and a hidden Markov model (i) to assess hazardous places of a contagious disease, and (ii) to predict the probability of individuals’ infection. To this end, we track the trajectories of individuals in an environment. For evaluating the models and the approaches, we use the Covid-19 outbreak in an urban environment as a case study. Individuals in a closed population are explicitly represented by their movement trajectories over a period of time. The simulation results demonstrate that by adjusting the communicable disease parameters, the detector system and the predictor system are able to correctly assess the hazardous places and determine the infection possibility of individuals and cluster them accurately with high probability, i.e., on average more than 96%. In general, the proposed approaches to assessing hazardous places and predicting the infection possibility of individuals can be applied to contagious diseases by tailoring them to the influential features of the disease.
... Davoodi et al. [25] addressed the constraints associated with conventional methodologies by proposing the integration of learning-based systems to transform hazard assessment and individual patient diagnoses. The main objective of this initiative is to address current disparities in healthcare practices through the integration of data-driven approaches, leading to enhanced precision in risk forecasting and significantly contributing to the progress of customized treatments. ...
... The average observation/person and observation intervals are the variants for data validation. The methods GA-GRU (genetic algorithm-based gated recurrent unit) [25], GVT (greedy-Voronoi tessellation) [29], and PCovNet (pre-symptomatic COVID-19 detection framework) [21] were used alongside the proposed CRS-PH method in the comparative analysis. ...
The recent impact of COVID-19, as a contagious disease, led researchers to focus on designing and fabricating personal healthcare devices and systems. With the help of wearable sensors, sensing and communication technologies, and recommendation modules, personal healthcare systems were designed for ease of use. More specifically, personal healthcare systems were designed to provide recommendations for maintaining a safe distance and avoiding contagious disease spread after the COVID-19 pandemic. The personal recommendations are analyzed based on the wearable sensor signals and their consistency in sensing. This consistency varies with human movements or other activities that hike/cease the sensor values abruptly for a short period. Therefore, a consistency-focused recommendation system (CRS) for personal healthcare (PH) was designed in this research. The hardware sensing intervals for the system are calibrated per the conventional specifications from which abrupt changes can be observed. The changes are analyzed for their saturation and fluctuations observed from neighbors within the threshold distance. The saturation and fluctuation classifications are performed using random forest learning to differentiate the above data from the previously sensed healthy data. In this process, the saturated data and consistency data provide safety recommendations for the moving user. The consistency is verified for a series of intervals for the fluctuating sensed data. This alerts the user if the threshold distance for a contagious disease is violated. The proposed system was validated using a prototype model and experimental analysis through false rates, data analysis rates, and fluctuations.
... The Markov model, a probabilistic prediction model based on statistics, utilizes real-time data to predict the number of affected and recovered patients, as well as the death count [8] [9]. This model constructs a probability transfer matrix to analyze the progression of the virus and make predictions based on the patterns observed in the data [10]. By employing the Markov model, researchers can gain insights into the future trajectory of the pandemic and make informed decisions regarding public health measures. ...
... They worked with medical professionals. The database includes 3616 COVID-19 positive cases, 10,192 Normal,6012 Lung Opacity (Non-COVID lung infection), and 1345 Viral Pneumonia images. This dataset is composed of four classes defined as follows [18]: ...
Markov chains are probabilistic models that are useful in different image processing tasks. This paper presents an approach based on Hidden Markov Chain (HMCs) to diagnose COVID-19, that we named HMC-COVID-19. To assess the performance of the proposed solution, we use a public dataset of chest X-RAY images. Our solution has been evaluated using the accuracy of prediction. The preliminary results are promising and can be further enhanced.
In the modern medical sector, deep learning methods are proving exceptionally effective in enhancing disease detection and classification accuracy through the analysis of medical imagery. The paper introduces an innovative framework for identifying infectious diseases using advanced convolutional neural network (CNN) models. The dataset encompasses diverse images of diseases like pneumonia, COVID-19, lung opacity, and MERS which have been sourced from various repositories. These images are transformed into grayscale and subjected to data augmentation techniques like horizontal and vertical flipping. The pre-processed images are then employed to train the models like VGG16, ResNet152V2, DenseNet169, and MobileNetV2. Through comprehensive evaluation using metrics such as loss, accuracy, recall, precision, and F1 score, the paper reveals that MobileNetV2 stands out by attaining remarkable accuracy and recall rates of 88.09% and 88.56%, respectively, in detecting and classifying infectious diseases. The model's potential in assisting medical practitioners with diagnoses and interventions is thereby underscored.
COVID-19 has become the largest pandemic in recent history to sweep the world. This study is devoted to developing and investigating three models of the COVID-19 epidemic process based on statistical machine learning and the evaluation of the results of their forecasting. The models developed are based on Random Forest, K-Nearest Neighbors, and Gradient Boosting methods. The models were studied for the adequacy and accuracy of predictive incidence for 3, 7, 10, 14, 21, and 30 days. The study used data on new cases of COVID-19 in Germany, Japan, South Korea, and Ukraine. These countries are selected because they have different dynamics of the COVID-19 epidemic process, and their governments have applied various control measures to contain the pandemic. The simulation results showed sufficient accuracy for practical use in the K-Nearest Neighbors and Gradient Boosting models. Public health agencies can use the models and their predictions to address various pandemic containment challenges. Such challenges are investigated depending on the duration of the constructed forecast.
Background
COVID-19 poses a severe threat to global human health, especially the USA, Brazil, and India cases continue to increase dynamically, which has a far-reaching impact on people's health, social activities, and the local economic situation.
Methods
The study proposed the ARIMA, SARIMA and Prophet models to predict daily new cases and cumulative confirmed cases in the USA, Brazil and India over the next 30 days based on the COVID-19 new confirmed cases and cumulative confirmed cases data set(May 1, 2020, and November 30, 2021) published by the official WHO, Three models were implemented in the R 4.1.1 software with forecast and prophet package. The performance of different models was evaluated by using root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE).
Results
Through the fitting and prediction of daily new case data, we reveal that the Prophet model has more advantages in the prediction of the COVID-19 of the USA, which could compose data components and capture periodic characteristics when the data changes significantly, while SARIMA is more likely to appear over-fitting in the USA. And the SARIMA model captured a seven-day period hidden in daily COVID-19 new cases from 3 countries. While in the prediction of new cumulative cases, the ARIMA model has a better ability to fit and predict the data with a positive growth trend in different countries(Brazil and India).
Conclusions
This study can shed light on understanding the outbreak trends and give an insight into the epidemiological control of these regions. Further, the prediction of the Prophet model showed sufficient accuracy in the daily COVID-19 new cases of the USA. The ARIMA model is suitable for predicting Brazil and India, which can help take precautions and policy formulation for this epidemic in other countries.
Massive molecular testing for COVID-19 has been pointed out as fundamental to moderate the spread of the pandemic. Pooling methods can enhance testing efficiency, but they are viable only at low incidences of the disease. We propose Smart Pooling, a machine learning method that uses clinical and sociodemographic data from patients to increase the efficiency of informed Dorfman testing for COVID-19 by arranging samples into all-negative pools. To do this, we ran an automated method to train numerous machine learning models on a retrospective dataset from more than 8000 patients tested for SARS-CoV-2 from April to July 2020 in Bogotá, Colombia. We estimated the efficiency gains of using the predictor to support Dorfman testing by simulating the outcome of tests. We also computed the attainable efficiency gains of non-adaptive pooling schemes mathematically. Moreover, we measured the false-negative error rates in detecting the ORF1ab and N genes of the virus in RT-qPCR dilutions. Finally, we presented the efficiency gains of using our proposed pooling scheme on proof-of-concept pooled tests. We believe Smart Pooling will be efficient for optimizing massive testing of SARS-CoV-2.
Finding the shortest path on uncertain transportation networks is a great challenge in theory and practice. There are several resources of uncertainty in the transportation networks such as traffic congestion, weather conditions, vehicle accidents, repairing roads, etc. A natural way to model uncertain networks is utilizing graphs with uncertain edges, that is, the weight of each edge, as the traveling time, may vary in an interval. In this paper, we not only discuss the theoretical aspect of this issue, but also propose a practical approach to find the shortest path on uncertain graphs. The approach has two phases; a preprocessing phase and a query phase. In the preprocessing phase, we construct a general map that contains all the edges that may lie on some shortest path, and in the query phase, when the weights are determined certainly, we find the shortest path using the map. Finally, we demonstrate the effectiveness of the proposed algorithm on both random generated graphs and real-world networks. Also, we compare it with other shortest path algorithms in the uncertainty context and shows its efficiency.
This work compares two innovative methodologies to predict future locations of vehicles when their current and previous locations are known. The two methodologies are based on a Bayesian network and a deep learning approach based on recurrent neural networks (RNNs). Several experimental results indicate that the prediction accuracy is improved as more information about prior vehicle mobility is available. The Bayesian method is advantageous because the statistical inference can be updated in real-time as more trajectory data is known. On the contrary, the RNN-based method requires a time-consuming learning task every time new data is added to the inference dataset. The results show that the computational cost to predict the next position a vehicle will move to can be substantially reduced when the Bayesian network is adopted. But when the quantity of prior data used in the prediction increases, the computational time required by the RNN-based method can be two orders of magnitude lower, showing that the RNN method is advantageous in both accuracy and computational time. Both methods achieve a next position successful prediction rate higher than 90%, confirming the applicability and validity of the proposed methods.
Deep learning is a powerful technique which is inspired by the structure as well as processing power of the human brain. This technique uses deep neural network to perform complex tasks such as time series prediction, image classification, and cancer detection. In this research work, we used Covid-19 time series datasets and with the help of deep learning we built the model for prediction of Covid-19 cases. For the model building, we used two deep learning neural networks, Recurrent Neural Networks (RNN) and Long Short Term Memory Networks (LSTMs). We built a prediction model using RNN in the first instance and subsequently the second model was built using LSTMs. Out of these two neural networks, we got promising results from the model based on LSTMs with an overall accuracy of 98%. As the cases of Covid-19 are increasing day-by-day at a very high rate, we proposed these models using neural networks to help in predicting the future trends of Covid-19 confirmed, deaths and recovered cases.
Demand‐side management (DSM) enables customers to decide consciously on how to seek and obtain power from the grid. The prevailing method available in DSM is load shifting. The grid is assisted through reducing load demands during the peak hours and altering the demand time into the off‐peak hours in a manner that the consumption sources could be met and several online load alterations may be precluded. The present article develops three approaches based on centralized multiagent reinforcement learning (CMARL) wherein the grid is modeled via a cooperative game. In the proposed methods, learning takes place in a center agent, and agents, in turn, are not to communicate with each other throughout the learning process. The results of implementations indicate that the proposed producers optimize the grid performance through diminishing customer costs and enhancing security by minimizing inter‐customer interactions.
Suppressing SARS-CoV-2 will likely require the rapid identification and isolation of infected individuals on an ongoing basis. Reverse transcription polymerase chain reaction (RT-PCR) tests are accurate but costly, making regular testing of every individual expensive. The costs are a challenge for all countries and particularly for developing countries. Cost reductions can be achieved by pooling (or combining) subsamples and testing them in groups1–7. A balance must be struck between increasing the group size and retaining test sensitivity, since sample dilution increases the likelihood of false negatives for individuals with low viral load in the sampled region at the time of the test⁸. Likewise, minimising the number of tests to reduce costs must be balanced against minimising the time testing takes to reduce the spread of infection. Here we propose an algorithm for pooling subsamples based on the geometry of a hypercube that, at low prevalence, accurately identifies infected individuals in a small number of tests and rounds of testing. We discuss the optimal group size and explain why, given the highly infectious nature of the disease, largely parallel searches are preferred. We report proof of concept experiments in which a positive subsample was detected even when diluted 100-fold with negative subsamples (cf. 30-fold to 48-fold dilution in Refs. 9–11). We quantify the loss of sensitivity due to dilution and discuss how it may be mitigated by frequent re-testing of groups, for example. With the use of these methods, the cost of mass testing could be reduced by a large factor which, furthermore, increases as the prevalence falls. Field trials of our approach are under way in Rwanda and South Africa. The use of group testing on a massive scale to closely and continually monitor infection in a population, along with rapid and effective isolation of infected people, provides a promising pathway to the longterm control of COVID-19.
Background:
The rate at which COVID-19 has spread throughout the globe has been alarming. While the role of fomite transmission is not yet fully understood, precise data on the environmental stability of SARS-CoV-2 is required to determine the risks of fomite transmission from contaminated surfaces.
Methods:
This study measured the survival rates of infectious SARS-CoV-2, suspended in a standard ASTM E2197 matrix, on several common surface types. All experiments were carried out in the dark, to negate any effects of UV light. Inoculated surfaces were incubated at 20 °C, 30 °C and 40 °C and sampled at various time points.
Results:
Survival rates of SARS-CoV-2 were determined at different temperatures and D-values, Z-values and half-life were calculated. We obtained half lives of between 1.7 and 2.7 days at 20 °C, reducing to a few hours when temperature was elevated to 40 °C. With initial viral loads broadly equivalent to the highest titres excreted by infectious patients, viable virus was isolated for up to 28 days at 20 °C from common surfaces such as glass, stainless steel and both paper and polymer banknotes. Conversely, infectious virus survived less than 24 h at 40 °C on some surfaces.
Conclusion:
These findings demonstrate SARS-CoV-2 can remain infectious for significantly longer time periods than generally considered possible. These results could be used to inform improved risk mitigation procedures to prevent the fomite spread of COVID-19.
COVID-19 is a novel virus that causes infection in both the upper respiratory tract and the lungs. The numbers of cases and deaths have increased on a daily basis on the scale of a global pandemic. Chest X-ray images have proven useful for monitoring various lung diseases and have recently been used to monitor the COVID-19 disease. In this paper, deep-learning-based approaches, namely deep feature extraction, fine-tuning of pretrained convolutional neural networks (CNN), and end-to-end training of a developed CNN model, have been used in order to classify COVID-19 and normal (healthy) chest X-ray images. For deep feature extraction, pretrained deep CNN models (ResNet18, ResNet50, ResNet101, VGG16, and VGG19) were used. For classification of the deep features, the Support Vector Machines (SVM) classifier was used with various kernel functions, namely Linear, Quadratic, Cubic, and Gaussian. The aforementioned pretrained deep CNN models were also used for the fine-tuning procedure. A new CNN model is proposed in this study with end-to-end training. A dataset containing 180 COVID-19 and 200 normal (healthy) chest X-ray images was used in the study's experimentation. Classification accuracy was used as the performance measurement of the study. The experimental works reveal that deep learning shows potential in the detection of COVID-19 based on chest X-ray images. The deep features extracted from the ResNet50 model and SVM classifier with the Linear kernel function produced a 94.7% accuracy score, which was the highest among all the obtained results. The achievement of the fine-tuned ResNet50 model was found to be 92.6%, whilst end-to-end training of the developed CNN model produced a 91.6% result. Various local texture descriptors and SVM classifications were also used for performance comparison with alternative deep approaches; the results of which showed the deep approaches to be quite efficient when compared to the local texture descriptors in the detection of COVID-19 based on chest X-ray images.