Machine Algorithm Research Articles

Machine learning-driven simplification of the hypomania checklist-32 for adolescent: a feature selection approach

BackgroundThe Hypomania Checklist-32 is widely used to screen for bipolar disorder, but its length can be challenging for adolescents with manic symptoms. This study aimed to develop a shortened version of the HCL-32 tailored for adolescents using machine learning techniques.MethodsData from 2,850 adolescents (mean age 15.50 years, 68.81% female) who completed the HCL-32 were analyzed. Random forest (RF) and gradient boosting machine (GBM) algorithms were employed for feature selection. The area under the curve (AUC) was used to evaluate model performance. Receiver operating characteristic (ROC) analysis was conducted to determine optimal cutoff points for the shortened scale.ResultsAn 8-item version of the HCL-32 was derived, maintaining high predictive accuracy (AUC = 0.97). The selected items captured core symptoms of adolescent mania, including increased energy, risk-taking, and irritability. Two cutoff points were identified: a score of 3 offered high specificity (0.98) and positive predictive value (0.98), while a score of 4 provided balanced sensitivity (0.87) and specificity (0.94) with the highest overall accuracy (0.91).ConclusionsThe machine learning-driven 8-item version of the HCL-32 demonstrates strong diagnostic utility for adolescent bipolar disorder, offering a more efficient screening tool without sacrificing clinical sensitivity. This shortened scale may improve assessment feasibility and accuracy in clinical settings, addressing the unique challenges of diagnosing bipolar disorder in adolescents.

Auxiliary identification of depression patients using interpretable machine learning models based on heart rate variability: a retrospective study

ObjectiveDepression has emerged as a global public health concern with high incidence and disability rates, which are timely imperative to identify and intervene in clinical practice. The objective of this study was to explore the association between heart rate variability (HRV) and depression, with the aim of establishing and validating machine learning models for the auxiliary diagnosis of depression.MethodsThe data of 465 outpatients from the Affiliated Hospital of Southwest Medical University were selected for the study. The study population was then randomly divided into training and test sets in a 7:3 ratio. Logistic regression (LR), support vector machine (SVM), random forest (RF) and eXtreme gradient boosting (XGBoost) algorithm models were used to construct risk prediction models in the training set, and the model performance was verified in the test set. The four models were evaluated by the area under the receiver operating characteristic curve (ROC), calibration curve and the decision curve analysis (DCA). Furthermore, we employed the SHapley Additive exPlanations (SHAP) method to illustrate the effects of the features attributed to the model.ResultsThere were 237 people in the depressed group and 228 in the non-depressed group. In the training set (n = 325) and test set (n = 140), the area under of the curve(AUC) values of the XGBoost model are 0.92 [95% confidence interval (CI) 0.888,0.95] and 0.82 (95% CI 0.754,0.892)] respectively, which are higher than the other three models. The XGBoost model has excellent predictive efficacy and clinical utility. The SHAP method was ranked according to the importance of the degree of influence on the model, with age, heart rate, Standard deviation of the NN intervals (SDNN), two nonlinear parameters of HRV and sex considered to be the top 6 predictors.ConclusionWe provided a feasibility study of HRV as a potential biomarker for depression. The proposed model based on HRV provides clinicians with a quantitative auxiliary diagnostic tool, which is assist to improving the accuracy and efficiency of depression diagnosis, and can also be utilized for the monitoring and prevention of depression.

Soft sensor modeling method for Pichia pastoris fermentation process based on substructure domain transfer learning

BackgroundAiming at the problem that traditional transfer methods are prone to lose data information in the overall domain-level transfer, and it is difficult to achieve the perfect match between source and target domains, thus reducing the accuracy of the soft sensor model.MethodsThis paper proposes a soft sensor modeling method based on the transfer modeling framework of substructure domain. Firstly, the Gaussian mixture model clustering algorithm is used to extract local information, cluster the source and target domains into multiple substructure domains, and adaptively weight the substructure domains according to the distances between the sub-source domains and sub-target domains. Secondly, the optimal subspace domain adaptation method integrating multiple metrics is used to obtain the optimal projection matrices Ws and Wt that are coupled with each other, and the data of source and target domains are projected to the corresponding subspace to perform spatial alignment, so as to reduce the discrepancy between the sample data of different working conditions. Finally, based on the source and target domain data after substructure domain adaptation, the least squares support vector machine algorithm is used to establish the prediction model.ResultsTaking Pichia pastoris fermentation to produce inulinase as an example, the simulation results verify that the root mean square error of the proposed soft sensor model in predicting Pichia pastoris concentration and inulinase concentration is reduced by 48.7% and 54.9%, respectively.ConclusionThe proposed soft sensor modeling method can accurately predict Pichia pastoris concentration and inulinase concentration online under different working conditions, and has higher prediction accuracy than the traditional soft sensor modeling method.

Hybrid prediction models for assessing the Higher Education Institutions Performance in QS World Institution Rankings

Background Quality Education is one of the primary requirements for the best survival. Pursuing higher education in a highly reputed institutions makes much difference in shaping the career of the individual. As many ranking and accreditation boards for higher education institutions like NAAC is prevalent, World ranking distinguishes institution reputation globally. The QS World University Ranking is a vital gauge for learners, educators, and institutions all over the world, allowing them to analyze and compare the quality and reputation of higher education. Predicting these rankings is difficult due to data availability concerns and QS’s frequent methodology revisions. Subjectivity and narrow criteria in rankings hamper the assessment of university greatness even more. Machine learning, data scraping, model adaptability, algorithm reversal, and short-term predictions are some existing ways of dealing with these difficulties. In this research, a prediction model for assessing institution performance in the QS World institution Rankings is designed using hybrid machine learning algorithms and optimization techniques. Two algorithms surpass others in forecasting ranks, according to the analysis. These hybrid models improves prediction accuracy of QS world rankings by integrating data analysis with model optimization using particle swarm optimization and Tabu search method.

Prediction of Disease Stage by Machine Learning Classification Methods for Covid-19 Patients

Supervised machine learning classificitaion algorithms have been widely used in many fields in recent years. Especially, health is one of the most important areas where machine learning studies are carried out successfully. The aim of this study is to develop models that predict the disease stage of people who apply to hospital with the diagnosis of Covid-19. Inadequacies such as intensive care occupancy, insufficiency of beds, and shortage of respiratory equipment are among these problems, and this has left healthcare workers faced with the overwhelming burden of patients. Therefore, estimating the disease stages of Covid-19 patients at an early stage is of great importance. The data set used in the study includes the clinical and laboratory data of the patients during in their admission to the hospital.It has been tried to develop models that predict disease stage by using Logistic Regression, Random Forest and Support Vector Machine algorithms in the data set. The random forest model with 9 variables was the best performing model. With the models obtained, it will be ensured that the hospital management receives information in order to see the necessary treatment for low-risk or high-risk patients and to avoid medical system inadequacies.

Development of an Intrusion Detection System using ANOVA Feature Selection and Support Vector Machine Algorithms

Development of an Intrusion Detection System using ANOVA Feature Selection and Support Vector Machine Algorithms

Prediction of Anthocyanin Content in Purple-Leaf Lettuce Based on Spectral Features and Optimized Extreme Learning Machine Algorithm

Prediction of Anthocyanin Content in Purple-Leaf Lettuce Based on Spectral Features and Optimized Extreme Learning Machine Algorithm

Prediction of potential occurrence of historical objects with defensive function in Slovakia using machine learning approach

In this article, we aim at the prediction of possible locations of already defunct historical objects with a defensive function (HODFs) in Slovakia, which have not been found and documented so far, using three machine learning methods. Specifically, we used the support vector machine, k-nearest neighbors, and random forest algorithms, which were trained based on the following five factors influencing the possible occurrence of HODFs: elevation, distance from a river, distance from a settlement, lithological rock type, and type of representative geoecosystems. Training and testing datasets were based on a database of already documented 605 HODFs, which were divided into 70% of training samples and 30% of testing samples. All of the three models reached the AUC-ROC value over 0.74 based on the testing dataset. The best performance was recorded by the random forest predictive model with the AUC-ROC value equal to 0.79. The results of the random forest model were also validated with the recently documented HODFs via the archeological research.

Exploring the role of alternative lengthening of telomere-related genes in diagnostic modeling for non-alcoholic fatty liver disease

Previous studies have reported an association between telomere length and non-alcoholic fatty liver disease (NAFLD). This study aimed to explore the involvement of alternative lengthening of telomere-related genes (ALTRGs) in the pathology of NAFLD, construct a risk signature, and evaluate both treatment and prognosis. Three NAFLD datasets (GSE48452, GSE89632, and GSE63067) were collected from the GEO database and merged into combined GEO datasets. ALTRGs were collected from GeneCards and PubMed databases. Differentially expressed genes (DEGs) were identified, and functional enrichment analysis was performed. This study employed a support vector machine algorithm and least absolute shrinkage and selection operator regression analysis to identify key genes for constructing a diagnostic model. High- and low-risk groups were identified from the combined GEO datasets using the diagnostic model. Gene set enrichment analysis, regulatory network analysis, and intergroup immune infiltration analysis were performed. This study identified the key genes using receiver operating characteristic and Friends analysis. Expression of these genes was validated in a mouse model of NAFLD. Twenty-five genes were differentially expressed, with a positive correlation between FOS and EGR1 and a negative correlation between MYC and CEBPA. A diagnostic model was constructed using 12 genes, and high- and low-risk groups were identified. CAMK2G, ERBB2, FOSB, WT1, and CEBPA showed certain accuracy, and their expression levels were significantly different in the model. Immune infiltration analysis between the risk groups revealed that six immune cells were statistically significant. This includes a strong negative interaction between type 2 T helper cells and SPHK2 in the high-risk group. These findings suggest that ALTRDEGs are potential therapeutic targets and prognostic indicators for NAFLD. However, further investigations are required to elucidate the specific underlying mechanisms.

Algorithms for graph theory matching problems based on computer science

Existing graph-matching algorithms only use the graph’s edges to transmit information and cannot perform hierarchical reasoning and fusion. Although they have achieved good results in graph matching, problems exist such as planarization and insufficient representation of graph data. This paper aims to study graph theory matching algorithms based on computer science, solve matching problems on large-scale graphs, reduce computation time, and improve solving efficiency. This paper studied graph theory matching algorithms based on computer science, including heuristic algorithm-based matching algorithms and machine learning-based matching algorithms; the specific matching problem was mathematically analyzed and transformed into a matching problem in graph theory, which included defining the input and output of the problem, determining the structure and properties of nodes and edges in the graph, and clarifying the required matching and constraint conditions. The experimental results of this paper showed that the average optimal matching rates of the Kuhn–Munkres algorithm, Gale–Shapley (GS)-based algorithm, Genetic Algorithm (GA), Particle Swarm Optimization (PSO) algorithm and Support Vector Machine (SVM) algorithm in complex scenarios were 62.05%, 58.42%, 82.79%, 81.99% and 83.50%, respectively. Studying graph theory matching algorithms based on computer science is of great significance. It can not only solve practical application problems and improve computational efficiency, but also promote algorithm optimization, which helps to improve the processing speed and efficiency of practical problems.

Classification model for blast furnace status based on multi-source information

The blast furnace (BF) is the key equipment for smelting, which has a significant impact on the sustainable development of global environment and energy. Therefore, it will be helpful for the operators if accurate evaluation to the status of the BF is realized. The classification model based on multi-source information is proposed in this paper, where the kernel fisher (KF) algorithm is enhanced to simplify the complexity both in algorithm and data dimensionality. The proposed model can realize the classification of both the distribution of gas flow (GF) and the geometric information of the burden surface (BS), and this paper also proposes a dynamic process adaptive kernel fisher (DPAKF) algorithm to enhance the adaptive ability of the classifier, where the online algorithm updating mechanism is designed. The experimental results demonstrate that the classification accuracy was increased from 90% to 95%, and the time expense of DPAKF was 236 s, which is prior compared with the existing support vector machine (SVM), k-nearest neighbor (KNN) and genetic algorithm (GA) models. The results of the student's t-distribution show that the statistics between DPAKF and KF, SVM, KNN and GA were 1.25, 2.81, 2.84 and 2.96, respectively, which demonstrate that there are significant statistical differences between DPAKF and SVM, KNN, and GA. This research provides a potential solution to the classification problem of the BF.

Development and Field Deployment of a ppb-Level SO2/NO2 Dual-Gas Sensor System for Agricultural Early Fire Identification.

Sulfur dioxide (SO2) and nitrogen dioxide (NO2) are chemical indicators of crop straw combustion as well as significant atmospheric pollutants. It is challenging to promptly detect natural "wildfires" during agricultural production, which often lead to uncontrollable and substantial economic losses. Moreover, both "wildfires" and artificial "straw burning" practices pose severe threats to the ecological environment and human health. Consequently, developing sensors capable of rapid and high-precision quantitative analysis of SO2/NO2 is essential and urgent for detecting early fires in agricultural activities. Here, we demonstrate an incoherent broadband cavity-enhanced absorption spectroscopy (IBBCEAS) sensing system utilizing a 366 nm ultraviolet light emitting diode, designed for real-time, high-precision monitoring of SO2 and NO2 and is used for early fire detection validation. The optical resonant cavity is constructed within a 60 mm cage system mechanical structure, achieving a maximum optical path length of nearly 2 km with a length of ∼460 mm. The output light carrying information about the species and concentration of the analyte molecules is coupled into the miniaturized grating spectrometer via a fiber, and continuous spectral fitting and concentration inversion are performed on the computer. We propose a spectral analysis and concentration inversion model based on an improved particle swarm optimization-support vector machine (IPSO-SVM) algorithm. By discrimination of the absorption spectral characteristics of SO2/NO2, we achieve superior prediction accuracy. Experimental results indicate that the detection limits of SO2 and NO2 under the optimized averaging time are 77.5 parts per billion by volume (ppbv) and 0.037 ppbv, respectively. The field deployment of the sensor in scenarios such as continuous outdoor air pollution monitoring, in situ combustion feature identification, and early fire mobile detection has demonstrated the superior reliability and sensitivity of this sensor system.

Probabilistic-based identification of gunshot residues (GSR) using Laser-Induced Breakdown Spectroscopy (LIBS) and Support Vector Machine (SVM) algorithm

Firearm violence results in thousands of victims annually in Brazil, which is ascribed, among other factors, to the low homicide resolution rates. One method for generating forensic evidence in firearm discharges is the identification of gunshot residues (GSR), typically composed of inorganic particles containing primarily lead (Pb), barium (Ba), and antimony (Sb). The “gold standard” technique for GSR identification is Scanning Electron Microscopy with Energy-Dispersive X-ray Spectroscopy (SEM-EDS) which is highly accurate but demands significant time and labor for each sample analysis. Other tests to detect GSR include the colorimetric rhodizonate test and spectroscopic techniques such as Laser-Induced Breakdown Spectroscopy (LIBS), which are quicker, but exhibit lower reliability and a higher potential for false positives and potentially compromise the investigation. This study introduces a novel GSR analytical protocol as a potential alternative to SEM-EDS and colorimetric tests, which is based on LIBS analysis combined with dimensionality reduction performed by PCA and probabilistic Support Vector Machine (SVM) algorithm. The developed protocol accurately predicts both positive and negative samples, and establishes the probability of each prediction, allowing for the exclusion of questionable samples from the analysis. The initial tests show the protocol achieved 100 % accuracy in both training and external validation. As a stress test, the protocol is able to exclude questionable samples from the analysis and accurately predict true samples with 100 % correct classifications from a set of artificially contaminated samples containing elements characteristic of GSR, highlighting the protocol’s significant reduction in the likelihood of false positives.

Model GHT-SVM for Traffic Sign Detection Using Support Vector Machine Algorithm Based On Gabor Filter and Top-Black Hat Transform

A factor that can hinder the detection and recognition of traffic signs is the variation in lighting in the image of traffic signs. This study aims to detect traffic symbols using Gabor Filter (GFT), Top Hat Transform (THT), and Black Hat Transform (BHT) methods on the Support Vector Machine (SVM) algorithm for traffic sign dataset images with data problems that tend to have dark backgrounds at night and bright backgrounds during the day. From the experimental results, GHT-SVM gets the highest accuracy compared to HSV-SVM, HSV-RF, HSV-KNN, and H2T-SVM models. Based on experimental results, H2T-SVM from HOG ⊕ ENT ⊕ BHT ⊕ SVM results get the best accuracy of 86.42%. The Gabor Filter (GFT) parameters used are the number of filters with a value of 16, ksize with a value of 30, sigma with a standard deviation value of 3.0, lambd with a sinusoidal factor value of 10.0, gamma with a spatial aspect ratio value of 0.5 and psi with a phase offset value of 0 while the Top Hat Transform (THT) and Black Hat Transform (BHT) methods use filterSize sizes with values (3, 3).

Unsaturated Fatty Acid Profiles and Prognostic Significance in Epilepsy Patients: A Comprehensive Analysis Using UPLC-MS/MS and SVM Algorithm

Unsaturated fatty acids (UFAs) play a crucial physiological role in human body. However, the concentration-related changes and prognostic significance of UFAs in epilepsyremain unclear. An optimized ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) approach was developed to measure six key UFAs: oleic acid (OA), linoleic acid (LA), arachidonic acid (AA), α-linolenic acid (ALA), docosahexaenoic acid (DHA), and eicosapentaenoic acid (EPA). Subsequently, the levels of these six UFAs were determined in 40 healthy individuals and 49 epilepsy patients. The diagnostic value of UFAs and clinical examination indicators were assessed using statistical analysis and the support vector machine (SVM) algorithm. The results showed that the UPLC-MS/MS method successfully quantified the levels of OA, LA, AA, ALA, EPA, and DHA in both the healthy individuals and epilepsy patients. Compared with the healthy group, the levels of ALA, AA, and DHA were significantly elevated in the epilepsy group (P < 0.05). Pearson correlation analysis revealed a strong positive correlation among the UFAs in the epilepsy group. The orthogonal partial least squares-discriminant analysis (OPLS-DA) model showed that DHA and EPA were more important than cholesterol in distinguishing between two groups, although the separation was not complete. The SVM model achieved better separation, with an area under the curve (AUC) of 0.95 when including the six UFAs. The EPA/DHA ratio was identified as a key feature, with a significant contribution to the model's performance. Removing the six UFAs from the model reduced the AUC to 0.91, highlighting the predictive value of UFAs for epilepsy. In conclusion, ALA, AA, and DHA, are altered in epilepsy patients. The EPA/DHA ratio was found to be a key predictive indicator for epilepsy. The use of UFAs in conjunction with clinical examination data improved the predictive power of the SVM model, suggesting that UFAs have potential as biomarkers for epilepsy.

Free vibration behaviour of bio-inspired helicoidal laminated composite panels of revolution under thermal conditions: Multi-output machine learning approach

The present work aims to study the free vibration behaviour of bio-inspired helicoidal laminated composite spherical, toroid, and conical shell panels using a single-output Support Vector Machine (SVM) algorithm trained in the chassis of parabolic shear deformation theory under thermal conditions. Different helicoidal lamination schemes are adopted, such as Fibonacci, semi-circular, exponential, recursive, and linear helicoidal schemes. Temperature-dependent material properties are adopted. The effect of the geometry of the shell, temperature, and lamination scheme on the free vibration behaviour of spherical, toroid, and conical shell panels is studied. Also, the mode shapes are obtained using different multi-output SVM surrogate in which the displacements are obtained at different locations and are predicted to obtain the fundamental mode shape. The trained surrogate model can predict the values of fundamental frequency and mode shapes much faster than the parabolic shear deformation theory.

Hyperparameter recommendation via automated meta-feature selection embedded with kernel group Lasso learning

Hyperparameter recommendation via meta-learning relies on the characterization and quality of meta-features. These meta-features provide critical information about the underlying datasets but are often selected manually based on the practitioner’s experience and preference, which can be inefficient and ineffective in many applications. In this paper, we propose a novel hyperparameter recommendation approach that integrates with a Lasso-based multivariate kernel group (KGLasso) model. The developed KGLasso model automatically identifies primary meta-features through model training. By selecting the most explanatory meta-features for a specific meta-learning task, the recommendation performance becomes much more effective. Our KGLasso model builds on a group-wise generalized multivariate Lasso approach. Within this framework, we establish a minimization algorithm using a corresponding auxiliary function, which is mathematically proven to be convergent and robust. As an application, we develop a hyperparameter recommendation system using our built KGLasso model on 120 UCI datasets for the well-known support vector machine (SVM) algorithm. This system efficiently provides competent hyperparameter recommendations for new tasks. Extensive experiments, including comparisons with popular meta-learning baselines and search algorithms, demonstrate the superiority of our proposed approach. Our results highlight the benefits of integrating model learning and feature selection to construct an automated meta-learner for hyperparameter recommendation in meta-learning.

Machine learning aided prediction of martensite transformation temperature of NiTi-based shape memory alloy

The martensitic transformation temperature (Ms) of shape memory alloys plays a crucial role in their performance. To achieve rapid prediction of Ms in shape memory alloys, this study employs machine learning techniques, using NiTi-based SMAs, the most widely used type, as an example. The results show that the gradient boosting machine algorithm provides the best prediction accuracy, with an R² of 0.92 and a mean absolute error (MAE) of 23.42 °C on the test set. Through correlation analysis, recursive feature elimination, and exhaustive search methods, six key features were identified. Subsequently, SHapley Additive exPlanations (SHAP) values were introduced to analyze the model interpretability, revealing the feature importance ranking and the dependence relationship between features and SHAP values. To address the issue of low Ms in NiTi-based shape memory alloys, the concept of high-entropy alloys was introduced. The study designed a TiZrHfNiCu high-entropy shape memory alloy and efficiently screened alloy compositions using the predictive model. This approach successfully increased the Ms, with the Ti19Zr19Hf19Ni37Cu6 alloy exhibiting an Ms exceeding 400 °C, significantly enhancing high-temperature application performance.

Intuitionistic Fuzzy SVM based on Kernel Gray Relational Analysis

Fuzzy Support Vector Machine (FSVM) is a machine learning algorithm that combines fuzzy logic with Support Vector Machine (SVM) to deal with the uncertainty and fuzziness in classification and regression problems. This algorithm improves the performance of traditional SVM by introducing fuzzy membership degrees, making it more robust when handling datasets with noise or uncertainty. Although the existing FSVM algorithms can overcome the influence of noise to a certain extent, they cannot effectively distinguish outliers or abnormal values from boundary support vectors. To solve this problem, this study proposes an Intuitionistic Fuzzy Support Vector Machine algorithm (KGRA-IFSVM) based on Kernel Grey Relational Analysis (KGRA). This approach utilizes gray relational analysis in the kernel space to calculate the gray relational degree between each sample and its K isomorphic neighboring points, and takes the average value as the membership degree of the sample. Then, the same approach is used to compute the gray relational degree between each sample and its K heterogeneous neighboring points, and the average value is taken as the non-membership degree of the sample. Finally, each sample is assigned with an appropriate fuzzy value based on intuitionistic fuzzy sets using a specific scoring function. Test results on UCI datasets show that KGRA-IFSVM has better classification performance and stronger noise resistance.

ACO-Guided Genetic Analysis: Exploring the Role of Genes in Alzheimer's Disease

Alzheimer's disease (AD), a neurodegenerative disorder significantly impairing cognitive function and quality of life, necessitates an in-depth study of its pathogenic mechanisms. Despite extensive research, the disease's pathological mechanisms harbor numerous unknown elements. Analyzing DNA microarray data is crucial for elucidating gene co-expression relationships during AD's pathological processes. This study aimed at exploring the applicability of the ant colony optimization (ACO) algorithm in identifying linearly co-expressed genes in AD. This study utilized gene expression profile data of AD patients collected from the Gene Expression Omnibus (GEO) database, employing the ACO algorithm to explore the co-expression relationships among AD pathogenic genes in different stages. Through the analysis of co-expression relationships, we identified four related genes (SDHA, NDUFA10, SDHC, and GPI). By using the co-expressed genes as features, we applied the support vector machine (SVM) algorithm for AD classification. The results of these experiments revealed that each model achieved average AUC values of 0.90, 0.91, 0.86, and 0.92, respectively. Our research findings reveal that the ACO algorithm provides a new perspective for in-depth exploration of the pathogenic mechanisms of AD.