Machine Learning and Deep Learning Approaches for Predicting Diabetes Progression: A Comparative Analysis

Introduction While the majority of intracranial aneurysms (IAs) remain asymptomatic, rupture results in subarachnoid hemorrhage, a devastating event with high morbidity and mortality. Traditional rupture risk assessment relies on morphology, size, and location, yet these parameters lack individualized precision. Recent advances in machine learning (ML) and radiomics provide an opportunity to integrate high‐dimensional imaging features and clinical data to predict rupture risk with greater accuracy. This systematic review and meta‐analysis evaluates the performance of ML‐ and radiomics‐based approaches for IA rupture prediction. Methods We reviewed studies applying ML or radiomics for IA rupture risk prediction, including systematic reviews, retrospective datasets, and prospective validation studies. Eligible reports used angiographic imaging (CTA, MRA, or DSA) with reference standards of rupture status or clinical outcomes. Extracted data included sample size, sensitivity, specificity, and area under the curve (AUC). Random‐effects models were used to synthesize predictive performance. Subgroup analyses evaluated conventional ML algorithms, radiomics‐based models, and deep learning approaches. Results Four representative studies highlight the evolving field. A meta‐analysis by Akhtar et al. (2024) demonstrated pooled sensitivity and specificity of 83% for ML rupture prediction, with AUC ∼0.85. Stumpo et al. (2022) reported that ML models consistently outperformed conventional risk scores, achieving sensitivity and specificity near 80%. Hu et al. (2023), using digital subtraction angiography in 263 patients, achieved exceptional performance with an AUC of 0.982, sensitivity of 94.4%, and specificity of 97.5% in distinguishing ruptured from unruptured aneurysms. In contrast, Heo et al. (2020) leveraged a national claims dataset to develop population‐level ML models, reaching AUROC 0.765 for rupture prediction, demonstrating the feasibility of integrating epidemiological and imaging data. Pooled analysis across studies yielded an average AUC of 0.85 (95% CI: 0.80‐0.89), with sensitivity ∼0.85 and specificity ∼0.86. Notably, radiomics models incorporating texture, intensity, and morphological signatures demonstrated superior discrimination compared to size‐based parameters alone. However, performance was heterogeneous across datasets, and generalizability was limited by small sample sizes and retrospective designs. Conclusion Machine learning and radiomics approaches extend aneurysm assessment “beyond detection,” offering promising accuracy in rupture risk prediction. With pooled AUCs near 0.85 and select studies reporting >0.95 performance, these models surpass traditional size‐based criteria. However, variability across cohorts and lack of standardized imaging features limit clinical translation. Future directions include prospective multicenter validation, integration of radiomics with clinical and hemodynamic markers, and development of explainable AI frameworks to ensure reproducibility. Ultimately, ML‐guided rupture risk stratification holds potential to personalize management and reduce subarachnoid hemorrhage incidence.

Considerable attention has recently been focused on classification and regression-based convolutional neural network (CNN) and long short-term memory (LSTM) due to their excellent performance in capturing complex spatial and temporal information characteristics for structural damage identification. However, few studies have considered structural damage identification as a classification and regression problem. In addition, bridges in practical engineering are vulnerable to various environmental and vehicle loading conditions. Hence, this study proposed a new two-stage CNN–LSTM configuration for bridge damage identification using vibration data considering the influence of temperatures. First, a classification-based CNN–LSTM is designed to perform multiclass damage detection tasks, and then a regression-based CNN–LSTM is developed for damage localization and severity prediction tasks. The performance of the proposed damage identification method was evaluated through a simulation dataset of a concrete highway bridge model and a field experiment dataset of Z24-bridge (Switzerland). In addition, a set of statistical evaluation metrics such as sparse categorical cross-entropy loss, accuracy, confusion matrix, mean squared loss, mean absolute error, mean absolute percentage error, and coefficient of determination were used to compare the damage identification performance of the proposed CNN–LSTM configuration with a regular CNN model and conventional machine learning (ML) algorithms. Prediction results indicate that the proposed CNN–LSTM model outperforms the regular CNN model and conventional ML algorithms for bridge damage identification.

Researchers have adopted model-centric artificial intelligence (AI) approaches in cheminformatics by using newer, more sophisticated AI methods to take advantage of growing chemical libraries. It has been shown that complex deep learning methods outperform conventional machine learning (ML) methods in QSAR and ligand-based virtual screening– but such approaches generally lack explanability. Hence, instead of developing more sophisticated AI methods (i.e., pursuing a model-centric approach), we wanted to explore the potential of a data-centric AI paradigm for virtual screening. A data-centric AI is an intelligent system that would automatically identify the right type of data to collect, clean and curate for later use by a predictive AI and this is required given the large volumes of chemical data that exist in chemical databases – PubChem alone has over 100 million unique compounds. However, a systematic assessment of the attributes and properties of suitable data is needed. We show here that it is not the result of deficiencies in current AI algorithms but rather, poor understanding and erroneous use of chemical data that ultimately leads to poor predictive performance. Using a new benchmark dataset of BRAF ligands that we developed, we show that our best performing predictive model can achieve an unprecedented accuracy of 99% with a conventional ML algorithm (SVM) using a merged molecular representation (Extended+ ECFP6 fingerprints), far surpassing past performances of virtual screening platforms using sophisticated deep learning methods. Thus, we demonstrate that it is not necessary to resort to the use of sophisticated deep learning algorithms for virtual screening because conventional ML can perform exceptionally well if given the right data and representation. We also show that the common use of decoys for training leads to high false positive rates and its use for testing will result in an over-optimistic estimation of a model’s predictive performance. Another common practice in virtual screening is defining compounds that are above a certain pharmacological threshold as inactives. Here, we show that the use of these so-called inactive compounds lowers a model’s sensitivity/recall. Considering that some target proteins have a limited number of known ligands, we wanted to also observe how the size and composition of the training data impact predictive performance. We found that an imbalance training dataset where inactives outnumber actives led to a decrease in recall but an increase in precision, regardless of the model or molecular representation used; and overall, we observed a decrease in the model’s accuracy. We highlight in this study some of the considerations that one needs to take into account in future development of data-centric AI for CADD.

Performance optimization of wireless networks is typically complicated because of high computational complexity and dynamic channel conditions. Considering a specific case, the recent introduction of intelligent reflecting surface (IRS) can reshape the wireless channels by controlling the scattering elements' phase shifts, namely, passive beamforming. However, due to the large size of scattering elements, the IRS's beamforming optimization becomes intractable. In this article, we focus on machine learning (ML) approaches for complex optimization problems in wireless networks. ML approaches can provide flexibility and robustness against uncertain and dynamic systems. However, practical challenges still remain due to slow convergence in offline training or online learning. This motivated us to design a novel optimization-driven ML framework that exploits the efficiency of model-based optimization and the robustness of model-free ML approaches. Splitting the control variables into two parts allows one part to be updated by the outer loop ML approach while the other part is solved by the inner loop optimization. The case study in IRS-assisted wireless networks confirms that the optimization-driven ML framework can improve learning efficiency and the reward performance significantly compared to conventional model-free ML approaches.

The introduction of various technologies in the context of Industry 4.0 allowed collecting monitoring data for various fields in manufacturing. Shop-floor and production data can be used for further analysis to extract knowledge. In this paper, an extensive evaluation of ten Machine Learning (ML) models for anomaly detection in manufacturing is conducted. The evaluation is conducted on multiple distinct ML algorithms, including conventional ML and a representative of Deep Neural Network (DNN) based algorithms. The ML models are trained on real production schedules to detect anomalous behavior in the overall system efficiency as well as violations in the delivery date of jobs. Multiple combinations of relevant features are tested during the training of the models. In essence, the objective of the ML models is to detect anomalous unknown breakdowns in the machines that lead to disruption in the overall performance of the system. The evaluation of the ML models is conducted on independent datasets with artificially injected machine breakdowns to establish the ground truth. These schedules with machine breakdowns are obtained through a simulation model. The results point to the high performance of a conventional ML algorithm, KNN. The performance demonstrated by a DNN-based AutoEncoder, conventional Local Outlier Factor-based Feature Bagging, and a recent Copula-Based Outlier Detection (COPOD) algorithms suggest limited applicability, which strongly depends on the shape of the data.

Early detection of hardware vulnerabilities is critical to ensuring the security and reliability of electronic systems. This study investigates and compares various machine learning and deep learning algorithms for their effectiveness in identifying hardware vulnerabilities during the early stages of the development lifecycle. The analysis focuses on algorithms commonly used in hardware security applications, including anomaly detection, classification, and pattern recognition, to detect issues such as hardware Trojans, side-channel attacks, and other security threats. Key performance metrics such as accuracy, detection speed, computational efficiency, and robustness against adversarial scenarios are evaluated. Furthermore, the study highlights the trade-offs between machine learning and deep learning approaches, considering their scalability and deployment feasibility in resource-constrained environments. The findings aim to provide insights into selecting the most suitable algorithm for early vulnerability detection, contributing to the advancement of secure hardware design and the mitigation of potential risks in critical systems.

Within the domain of password security classification, the pursuit of practical and dependable methodologies has prompted the examination of both biological and technological paradigms. The present study investigates the efficacy of Mouth Brooding Fish (MBF) as an innovative method in contrast to conventional Machine Learning (ML) approaches for classifying password security. The research approach entails a rigorous examination of the comparative analysis of MBF and ML algorithms, evaluating their effectiveness in password classification using many criteria, including accuracy, robustness, flexibility, and durability against adversarial assaults. The findings suggest that ML approaches have shown significant effectiveness in classifying passwords. However, using methodologies inspired by the minimum Bayes risk framework demonstrates a higher degree of resistance against typical cyber dangers. The intrinsic biological mechanisms of MBF, encompassing adaptive behaviors and inherent protection, play a role in enhancing the resilience and adaptability of the password security categorization system. The results offer significant insights that can inform the evolution of password security systems, integrating biological principles with technical progress to enhance safeguarding measures in digital environments. To emphasize the advantages of the suggested approach, several ML approaches are investigated, such as Support Vector Machines (SVM), AdaBoost, Multilayer Perceptron (MLP), Gaussian Kernel (GK), and Random Forest (RF). The F-score, accuracy, sensitivity, and specificity metrics for MBF exhibit noteworthy performance compared to the other selected models, with values of 100%.

The aim of this research is maximizing differentiation quality of skin neoplasms by Raman Spectroscopy (RS) and Autofluorescence (AF) using conventional Machine Learning (ML) algorithms (which means excluding Neural Networks). Thus, a basic task of this research consists of making and testing ML algorithm ensembles on Raman spectra, that were obtained in vivo in Samara Oncology Clinical Center. The data (spectra) have been obtained in a form of text files containing the identifier of a patient, as well as a Raman spectrum in the form of pair values – a wavelength and an appropriate value. All data (964 spectra) has been divided for two classes: Tumor and Skin. Further preprocessing of the input data and the analysis of models of machine learning for a problem of classification have been carried out. We used the following ML tools, namely Python 3.7.3, an open source ML libraries Scikit-learn v0.21.2, NumPy 1.17.0, and Pandas 1.0.1, IDE Anaconda Enterprise 5.3 and cloud service Google Colaboratory with an interactive environment Jupyter Notebook. Machine learning models that show a high accuracy result include Classification and Regression Tree (CART), Support Vector Classification (SVC), Logistic Regression (LR), K-nearest neighbors algorithm (KNN). These classifiers show high quality classification on standard parameters already. The Soft Voting Classification module was selected as an ensemble, that allows us to use several models of classifiers that are not similar to each other at once, combine them into one classifier. Results of this ensemble testing from these ML algorithms showed that the classification accuracy, unlike the best qualifier, has not been increased. However, the metrics of classification quality show that the model has become stabler and steady. Results: specificity - 93%, sensitivity - 88, harmonic mean between precision and recall (F1 score) - 90%. An analysis of validation and training curves indicated a small size of training data and, for some cases, a high complexity of the model, which led to a decrease in the classification accuracy.

ObjectivesTo develop and identify machine learning (ML) models using pretreatment 2-deoxy-2-[18F]fluoro-D-glucose ([18F]-FDG)-positron emission tomography (PET)-based radiomic features to differentiate benign from malignant parotid gland diseases (PGDs).Materials and methodsThis retrospective study included 62 patients with 63 PGDs who underwent pretreatment [18F]-FDG-PET/computed tomography (CT). The lesions were assigned to the training (n = 44) and testing (n = 19) cohorts. In total, 49 [18F]-FDG-PET-based radiomic features were utilized to differentiate benign from malignant PGDs using five different conventional ML algorithmic models (random forest, neural network, k-nearest neighbors, logistic regression, and support vector machine) and the deep learning (DL)-based ensemble ML model. In the training cohort, each conventional ML model was constructed using the five most important features selected by the recursive feature elimination method with the tenfold cross-validation and synthetic minority oversampling technique. The DL-based ensemble ML model was constructed using the five most important features of the bagging and multilayer stacking methods. The area under the receiver operating characteristic curves (AUCs) and accuracies were used to compare predictive performances.ResultsIn total, 24 benign and 39 malignant PGDs were identified. Metabolic tumor volume and four GLSZM features (GLSZM_ZSE, GLSZM_SZE, GLSZM_GLNU, and GLSZM_ZSNU) were the five most important radiomic features. All five features except GLSZM_SZE were significantly higher in malignant PGDs than in benign ones (each p < 0.05). The DL-based ensemble ML model had the best performing classifier in the training and testing cohorts (AUC = 1.000, accuracy = 1.000 vs AUC = 0.976, accuracy = 0.947).ConclusionsThe DL-based ensemble ML model using [18F]-FDG-PET-based radiomic features can be useful for differentiating benign from malignant PGDs.Second abstractThe DL-based ensemble ML model using [18F]-FDG-PET-based radiomic features can overcome the previously reported limitation of [18F]-FDG-PET/CT scan for differentiating benign from malignant PGDs. The DL-based ensemble ML approach using [18F]-FDG-PET-based radiomic features can provide useful information for managing PGD.

Objectives: Language detection is the process of identifying a language associated with a text. The proposed system aims to detect the Dravidian language that is associated with the given text using different machine learning and deep learning algorithms. The paper presents an empirical analysis of the results obtained using the different models. It also aims to evaluate the performance of a language agnostic model for the purpose of language detection. Method: An empirical analysis of Dravidian language identification in social media text using machine learning and deep learning approaches with k-fold cross validation has been implemented. The identification of Dravidian languages, including Tamil, Malayalam, Tamil Code Mix, and Malayalam Code Mix, is performed using both machine learning (ML) and deep learning algorithms. The machine learning algorithms used for language detection are Naive Bayes (NB), Multinomial Logistic Regression (MLR), Support Vector Machine (SVM), and Random Forest (RF). The supervised Deep Learning (DL) models used include BERT, mBERT and language agnostic models. Findings: The language agnostic model outperform all other models considering the task of language detection in Dravidian languages. The results of both the ML and DL models are analyzed empirically with performance measures like accuracy, precision, recall, and f1-score. The accuracy associated with different machine learning algorithms varies from 85% to 89%. It is evident from the experimental result that the deep learning model outperformed with an accuracy of 98%. Novelty: The proposed system emphasizes on the use of the language agnostic model to implement the process of detecting Dravidian languages associated with the given text which provides a promising result of 98% accuracy which is higher than the existing methodologies. Keywords: Language, Machine learning, Deep learning, Transformer model, Encoder, Decoder

This study investigates the feasibility of using tear sample analysis, based on protein corona formation on gold nanoparticles combined with electrospray ionization mass spectrometry (ESI-MS) and machine learning techniques, as a non-invasive approach for the detection of choroidal melanoma. The aim is to assess whether protein-nanoparticle interactions can support early and reliable identification of this ocular condition. Tear samples were collected using Schirmer strips from six healthy individuals and six patients diagnosed with choroidal melanoma, with subsequent augmentation to 18 samples per group. Gold nanoparticles (AuNPs, ~ 20 nm) were synthesized via citrate reduction and incubated with tear samples to form protein coronas, which were analyzed using ESI-MS. Eight statistical and entropy-based features (mean, variance, skewness, kurtosis, Shannon entropy, approximate entropy, sample entropy, and permutation entropy) were extracted from spectral data. Additionally, Continuous Wavelet Transform (CWT) with Mexican hat wavelet was applied to convert mass spectrometry data into 128 × 128 RGB images for deep learning analysis. Classification was performed using traditional machine learning models (Random Forest, Support Vector Machine, Decision Tree, Deep Neural Network) and transfer learning with pre-trained CNNs (VGG16, ResNet50, Xception), evaluated through 5-fold cross-validation. Significant differences in spectral intensity parameters were observed between healthy individuals and choroidal melanoma patients (p < 0.001), with notably lower Mean_Intensity values in cancer patients (56.41 ± 46.06 vs. 111.02 ± 10.01, Cohen’s d = 1.64). While m/z parameters showed moderate differences that didn’t reach statistical significance (p = 0.082), entropy-based features demonstrated strong discriminative power. Among traditional machine learning models, Random Forest achieved the highest accuracy (0.959 ± 0.003) and ROC AUC (0.993 ± 0.000) with remarkable computational efficiency (3.90 s per fold). For deep learning approaches using CWT-generated images, VGG16 demonstrated superior performance (Accuracy: 0.976 ± 0.008, ROC AUC: 0.997 ± 0.002) despite requiring significantly higher computational resources (1349.52 s per fold). This study demonstrates that tear sample analysis using protein corona formation on gold nanoparticles with ESI-MS and advanced machine learning techniques offers a promising non-invasive approach for choroidal melanoma detection with performance metrics that compare favorably to existing methods. The significant differences in spectral intensity parameters between groups suggest distinctive proteomic signatures that can be leveraged for diagnostic purposes. While both traditional machine learning and deep learning approaches achieved exceptional performance, each offers distinct advantages in terms of computational efficiency and feature extraction capabilities.

Identification of damage in its early stage can have a great contribution in decreasing the maintenance costs and prolonging the life of valuable structures. Although conventional damage detection techniques have a mature background, their widespread application in industrial practice is still missing. In recent years the application of Machine Learning (ML) algorithms have been more and more exploited in structural health monitoring systems (SHM). Because of the superior capabilities of ML approaches in recognizing and classifying available patterns in a dataset, they have demonstrated a significant improvement in traditional damage identification algorithms. This review study focuses on the use of machine learning (ML) approaches in Ultrasonic Guided Wave (UGW)-based SHM, in which a structure is continually monitored using permanent sensors. Accordingly, multiple steps required for performing damage detection through UGWs are stated. Moreover, it is outlined that the employment of ML techniques for UGW-based damage detection can be subtended into two main phases: (1) extracting features from the data set, and reducing the dimension of the data space, (2) processing the patterns for revealing patterns, and classification of instances. With this regard, the most frequent techniques for the realization of those two phases are elaborated. This study shows the great potential of ML algorithms to assist and enhance UGW-based damage detection algorithms.

Metabolic pathway modeling, essential for understanding organism metabolism, is pivotal in predicting genetic mutation effects, drug design, and biofuel development. Enhancing these modeling techniques is crucial for achieving greater prediction accuracy and reliability. However, the limited experimental data or the complexity of the pathway makes it challenging for researchers to predict phenotypes. Deep learning (DL) is known to perform better than other Machine Learning (ML) approaches if the right conditions are met (i.e., a large database and good choice of parameters). Here, we use a knowledge-based model to massively generate synthetic data and extend a small initial dataset of experimental values. The main objective is to assess if DL can perform at least as well as other ML approaches in flux prediction, using 68,950 instances. Two processing methods are used to generate DL models: cross-validation and repeated holdout evaluation. DL models predict the metabolic fluxes with high precision and slightly outperform the best-known ML approach (the Cubist model) with a lower RMSE (≤0.01) in both cases. They also outperform the PLS model (RMSE ≥ 30). This study is the first to use DL to predict the overall flux of a metabolic pathway only from variations of enzyme concentrations.

Deep learning (DL) is now generally acknowledged as the benchmark and evolution in machine learning (ML) fields. Further more, it has steadily suited the most extensively in computational techniques for ML, delivering excellent outcomes in several challenging intellectual works that have equal or still outperformed human ability. DL has the advantage of learning from enormous amounts of data. Here, the DL approach has been applied to classify thunderstorms and non-thunderstorms with a meteorological dataset. Daily observational hourly data sets from 2016 to 2021 have been used to classify the incidence of thunderstorms and non-thunderstorms. Three DL approaches are used: Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), and Long Short-Term Memory (LSTM) to classify thunderstorm incidences. These DL techniques are compared with conventional ML techniques, showing that DL approaches outperformed ML. The CNN method provides the highest performance with 92.01% accuracy among all DL and ML approaches.

There is an increasing need for artificial intelligence in the field of medical related learning frameworks which is capable of efficiently handling the multimodal data which is under resource constrained situations. Conventional machine learning and deep learning algorithms are usually depending on the large labelled dataset and limited environment with less applicability for real world applications. The adoption of the Artificial intelligence in the field of healthcare is the need for the learning and effective handling of the multimodal healthcare data. Traditional machine learning and deep learning algorithms depends on the large-scale datasets with label, limiting the usage of the current clinical environment. This will limit the applicability of the real-world clinical environments by the data lackness and variance in the domain. In this work, Large Language Models are used with deep metric learning framework for multimodal medical journal classification. In this work, Large Language Model is implemented using deep metric learning framework for the multimodal dataset. The multimodal dataset comprises of text, table and images from online repository. The data manually downloaded from the PubMed website is used in this work. The proposed LLM guide framework outperformed all other models trained by various benchmark datasets.