Artificial Intelligence-based Prediction Research Articles

Artificial Intelligence-Based Prediction of Covid-19 Severity on the Results of Protein Profiling

BackgroundCOVID-19 progresses slowly and negatively affects many people. However, mild to moderate symptoms develop in most infected people, who recover without hospitalization. Therefore, the development of early diagnosis and treatment strategies is essential. One of these methods is proteomic technology based on the blood protein profiling technique. This study aims to classify three COVID-19 positive patient groups (mild, severe, and critical) and a control group based on the blood protein profiling using deep learning (DL), random forest (RF), and gradient boosted trees (GBTs). MethodsThe dataset consists of 93 samples (60 COVID-19 patients, 33 control), and 370 variables obtained from an open-source website. The current dataset contains age, gender, and 368 protein, used to predict the relationship between disease severity and proteins using DL and machine learning approaches (RF, GBTs). An evolutionary algorithm tunes hyperparameters of the models and the predictions are assessed through accuracy, sensitivity, specificity, precision, F1 score, classification error, and kappa performance metrics. ResultsThe accuracy of RF (96.21%) was higher as compared to DL (94.73%). However, the ensemble classifier GBTs produced the highest accuracy (96.98%). TGB1BP2 in the cardiovascular II panel and MILR1 in the inflammation panel were the two most important proteins associated with disease severity. ConclusionsThe proposed model (GBTs) achieved the best prediction of disease severity based on the proteins compared to the other algorithms. The results point out that changes in blood proteins associated with the severity of COVID-19 may be used in monitoring and early diagnosis/treatment of the disease.

Application of artificial intelligence and machine learning for prediction of oral cancer risk.

Oral cancer requires early diagnosis and treatment to increase the chances of survival. This study aimed to develop an artificial neural network model that helps to predict the individuals' risk of developing oral cancer based on data on risk factors, systematic medical condition, and clinic-pathological features. A popular data mining algorithm artificial neural network was used for developing the artificial intelligence-based prediction model. A total of 29 variables that were associated with the patients were used for developing the model. The dataset was randomly split into the training dataset 54 (75%) cases and testing dataset 19 (25%) cases. All records and observations were reviewed by Board-certified oral pathologist. A total of 73 patients met the eligibility criteria. Twenty-two (30.13%) were benign cases, and 51 (69.86%) were malignant cases. Thirty-seven were female, and 36 were male, with a mean age of 63.09years. Our analysis displayed that the average sensitivity and specificity of ANN for oral cancer prediction based on the 10-fold cross-validation analysis was 85.71% (95% confidence interval [CI], 57.19-98.22) and 60.00% (95% CI, 14.66-94.73), respectively. The accuracy of ANN for oral cancer prediction was 78.95% (95% CI, 54.43-931.95). Our results suggest that this machine-learning technique has the potential to help in oral cancer screening and diagnosis based on the datasets. The results demonstrate that the artificial neural network could perform well in estimating the probability of malignancy and improve the positive predictive value that could help to predict the individuals' risk of developing OC based on knowledge of their risk factors, systemic medical conditions, and clinic-pathological data.

Artificial intelligence-based prediction of transfusion in the intensive care unit in patients with gastrointestinal bleeding

ObjectiveGastrointestinal (GI) bleeding commonly requires intensive care unit (ICU) in cases of potentialhaemodynamiccompromise or likely urgent intervention. However, manypatientsadmitted to the ICU stop bleeding and do not require further intervention,...

Development of shade prediction system to quantify the shade change after crease recovery finish application using artificial neural networks

Finishes are applied to improve the look, performance and feel of the fabrics. Crease recovery finishes form a three-dimensional crosslinking network on the surface of the cotton knitted fabric to control its dimensions. However, application of the crease recovery finishes induces the shade change in the dyed fabrics. This paper presents the phenomenon of shade change for different colors and shade percentages and use of artificial intelligence-based prediction system to foresee the behavior of shade after finish application. The individual neural networks were trained for the prediction of color of the finished samples, which are delta color coordinates values (△L, △a, △b, △c & △h). The input variables, i.e. reflectance values (Visible ranges 400–700 nm) of dyed samples, color, shade percentage and finish concentration were used to train the networks. The trained neural networks were validated through ‘cross validation’ and ‘hold out’ techniques. The shade prediction model was developed by combining the individually trained artificial neural networks and the developed model can predict the shade change with more than 90% accuracy. This will help the dyers to predict shade change prior to dyeing & finishing and they will adjust their recipe accordingly, which can ultimately reduce the rework and reprocessing in the textile wet processing industries.

Saltwater intrusion prediction in coastal aquifers utilizing a weighted-average heterogeneous ensemble of prediction models based on Dempster-Shafer theory of evidence

ABSTRACT Accurate and meaningful prediction of saltwater intrusion in coastal aquifers requires appropriate prediction tools. Artificial intelligence-based prediction models and their ensembles have been a better choice for mimicking the complex and nonlinear seawater intrusion progressions in coastal aquifers. This study utilizes a weighted-average ensemble of ‘heterogeneous’ prediction models to predict the saltwater intrusion progression in a coastal aquifer study area. The Dempster-Shafer theory of evidence is employed to calculate the weights of five different prediction model algorithms. Corresponding weights for individual prediction models are utilized in developing the ensemble prediction. Ensemble prediction performance for salinity intrusion in coastal aquifers in this effort is evaluated using several descriptive metrics. The values of the descriptive metrics suggest that the ensemble model performs in the same way as the best model in the ensemble. The methodology is evaluated for an illustrative coastal aquifer study area exposed to pumping-induced saltwater intrusion.

A Joint Evaluation of Impaired Cardiac Sympathetic Responses and Malnutrition-Inflammation Cachexia for Mortality Risks in Hemodialysis Patients.

Background: Cardiac sympathetic response (CSR) and malnutrition-inflammation syndrome (MIS) score are validated assessment tools for patients' health condition. We aim to evaluate the joint effect of CSR and MIS on all-cause and cardiovascular (CV) mortality in patients with hemodialysis (HD).Methods: Changes in normalized low frequency (ΔnLF) during HD were utilized for quantification of CSR. Unadjusted and adjusted hazard ratios (aHRs) of mortality risks were analyzed in different groups of ΔnLF and MIS score.Results: In multivariate analysis, higher ΔnLF was related to all-cause, CV and sudden cardiac deaths [aHR: 0.78 (95% confidence interval (CI): 0.72–0.85), 0.78 (95% CI: 0.70–0.87), and 0.74 (95% CI: 0.63–0.87), respectively]. Higher MIS score was associated with incremental risks of all-cause, CV and sudden cardiac deaths [aHR: 1.36 (95% CI: 1.13–1.63), 1.33 (95% CI: 1.06 – 1.38), and 1.50 (95% CI: 1.07–2.11), respectively]. Patients with combined lower ΔnLF (≤6.8 nu) and higher MIS score were at the greatest risk of all-cause and CV mortality [aHR: 5.64 (95% CI: 1.14–18.09) and 5.86 (95% CI: 1.64–13.65), respectively].Conclusion: Our data indicate a joint evaluation of CSR and MIS score to identify patients at high risk of death is more comprehensive and convincing. Considering the extremely high prevalence of cardiac autonomic neuropathy and malnutrition-inflammation cachexia in HD population, a non-invasive monitoring system composed of CSR analyzer and MIS score calculator should be developed in the artificial intelligence-based prediction of clinical events.

Predicting the principal strong ground motion parameters: A deep learning approach

In this research, new models are developed to estimate the three principal time-domain parameters of seismic ground motion. A novel deep learning (DL) approach coupled with artificial neural network (ANN), namely deep neural network (DNN) is employed for predicting the strong ground motion parameters such as peak ground acceleration (PGA), peak ground velocity (PGV) and peak ground displacement (PGD). This robust technique that has extended the applications of conventional neural networks improves learning of complicated and nonlinear features via increasing the number of layers as well as the neurons in each layer. The proposed models are constructed upon the NGA-West2 database provided by PEER (Pacific Earthquake Engineering Research Center). This database is more comprehensive than NGA-West1 which was mainly considered for developing previous artificial intelligence-based prediction models. Therefore, the new models are more reliable and can be used for wider ranges of predictors. The DNN attenuation models yield accurate estimates of the site PGA, PGV and PGD based on earthquake magnitude, rake angle, source to site distance and soil shear wave velocity. In addition, it is shown that the developed models, with correlation coefficients of 0.902, 0.899 and 0.911 (for PGA, PGV and PGD respectively), outperform the existing soft computing models. Furthermore, the average values of error measures such as MAE, MAPE and RMSE for PGA, PGV and PGD equal to 0.456, 0.758 and 0.581 compare favorably with those of previous models.

Electrical conductivity of ammonium and phosphonium based deep eutectic solvents: Measurements and artificial intelligence-based prediction

The evaluation of deep eutectic solvents (DESs) as a new generation of solvents for various practical application requires an insight of the main physical, chemical, and thermodynamic properties. In this study, the experimental measurements of the electrical conductivity of two classes of DESs based on ammonium and phosphonium salts at different compositions and temperatures were reported. The results revealed that electrical conductivity of DESs has temperature-dependency. In addition, molar conductivities of ammonium and phosphonium salts in DESs were obtained using DESs experimental values of electrical conductivities. The feasibility of using an artificial neural network (ANN) model to predict the electrical conductivity of ammonium and phosphonium based DESs at different temperatures and compositions was also examined. A feed-forward back propagation neural network with 8 hidden neurons was successfully developed and trained with the measured electrical conductivity data. The results indicated that among the different networks tested, the network with 8 hidden neurons had the best prediction performance and gave the smallest value of Normalized Mean Square Error (NMSE) (0.0010) and acceptable values of Index of Agreement (IA) (0.9999) and Regression Coefficient (R2) (0.9988). The comparison of the predicted electrical conductivity of DESs by the proposed model with those obtained by experiments confirmed the reliability of the ANN model with an average absolute relative deviation (AARD%) of 4.40%.

Protein structure prediction: Combining de novo modeling with sparse experimental data

Routine structure prediction of new folds is still a challenging task for computational biology. The challenge is not only in the proper determination of overall fold but also in building models of acceptable resolution, useful for modeling the drug interactions and protein-protein complexes. In this work we propose and test a comprehensive approach to protein structure modeling supported by sparse, and relatively easy to obtain, experimental data. We focus on chemical shift-based restraints from NMR, although other sparse restraints could be easily included. In particular, we demonstrate that combining the typical NMR software with artificial intelligence-based prediction of secondary structure enhances significantly the accuracy of the restraints for molecular modeling. The computational procedure is based on the reduced representation approach implemented in the CABS modeling software, which proved to be a versatile tool for protein structure prediction during the CASP (CASP stands for critical assessment of techniques for protein structure prediction) experiments (see http://predictioncenter/CASP6/org). The method is successfully tested on a small set of representative globular proteins of different size and topology, including the two CASP6 targets, for which the required NMR data already exist. The method is implemented in a semi-automated pipeline applicable to a large scale structural annotation of genomic data. Here, we limit the computations to relatively small set. This enabled, without a loss of generality, a detailed discussion of various factors determining accuracy of the proposed approach to the protein structure prediction.