Interpretation and Classification of Arrhythmia Using Deep Convolutional Network

XAI is critical for establishing trust and enabling the appropriate development of machine learning models. By offering transparency into how these models make judgements, XAI enables researchers and users to uncover potential biases, admit limits, and eventually enhance the fairness and dependability of AI systems. In this paper, we demonstrates two techniques, LIME and SHAP, used to improve the interpretability of machine learning models. Assessing Explainable AI (XAI) approaches is critical in searching for transparent and interpretable artificial intelligence (AI) models. Explainable AI (XAI) approaches are designed to provide insight into how complex models make decisions. This paper thoroughly analyzes two prominent XAI methods: Shapley Additive explanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME). This study aims to understand the decision made by a machine learning model and how the model came to that decision. We discuss the approaches and framework of both LIME and SHAP and assess their behavior in predicting the model's outcome.

Heart disease remains a major global health challenge, underscoring the need for predictive models that are both accurate and interpretable to support early diagnosis and clinical decision-making. Using Shapley Additive Explanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) in this study allowed for a deeper level of understanding of model predictions as well as provided viable alternative ways to overcome the challenges associated with black-box classifiers. Using a soft-voting ensemble technique that combined representations for Random Forest (RF), Decision Tree (DT), and Support Vector Machines (SVM) was applied to evaluate a publicly available heart disease dataset, containing 5110 records with each record containing 76 attributes. A total of 14 clinically relevant features were identified and evaluated using an 80/20 train-test split for the experimental setup, with results further validated by 5-fold cross-validation. The proposed soft-voting ensemble model achieved an accuracy of 89.10%, precision of 83.90%, recall of 95.20%, and F1-score of 89.90%, out-performing all individual baseline models and many of the recent XAI-enhanced deep learning models. The combination of SHAP and LIME allowed for both global and local interpretations of the model’s outputs, providing interpretable insights as to features like cp, thal, ca, and oldpeak contributed to the predictions and aligning these features with known cardiovascular risk factors. The results of this research demonstrate how interpretable (or explainable) ensemble-learning mechanisms can confer highly reliable, clinically useful heart disease risk predictions.

BackgroundTuberculosis spondylitis (TS), commonly known as Pott’s disease, is a severe type of skeletal tuberculosis that typically requires surgical treatment. However, this treatment option has led to an increase in healthcare costs due to prolonged hospital stays (PLOS). Therefore, identifying risk factors associated with extended PLOS is necessary. In this research, we intended to develop an interpretable machine learning model that could predict extended PLOS, which can provide valuable insights for treatments and a web-based application was implemented.MethodsWe obtained patient data from the spine surgery department at our hospital. Extended postoperative length of stay (PLOS) refers to a hospitalization duration equal to or exceeding the 75th percentile following spine surgery. To identify relevant variables, we employed several approaches, such as the least absolute shrinkage and selection operator (LASSO), recursive feature elimination (RFE) based on support vector machine classification (SVC), correlation analysis, and permutation importance value. Several models using implemented and some of them are ensembled using soft voting techniques. Models were constructed using grid search with nested cross-validation. The performance of each algorithm was assessed through various metrics, including the AUC value (area under the curve of receiver operating characteristics) and the Brier Score. Model interpretation involved utilizing methods such as Shapley additive explanations (SHAP), the Gini Impurity Index, permutation importance, and local interpretable model-agnostic explanations (LIME). Furthermore, to facilitate the practical application of the model, a web-based interface was developed and deployed.ResultsThe study included a cohort of 580 patients and 11 features include (CRP, transfusions, infusion volume, blood loss, X-ray bone bridge, X-ray osteophyte, CT-vertebral destruction, CT-paravertebral abscess, MRI-paravertebral abscess, MRI-epidural abscess, postoperative drainage) were selected. Most of the classifiers showed better performance, where the XGBoost model has a higher AUC value (0.86) and lower Brier Score (0.126). The XGBoost model was chosen as the optimal model. The results obtained from the calibration and decision curve analysis (DCA) plots demonstrate that XGBoost has achieved promising performance. After conducting tenfold cross-validation, the XGBoost model demonstrated a mean AUC of 0.85 ± 0.09. SHAP and LIME were used to display the variables’ contributions to the predicted value. The stacked bar plots indicated that infusion volume was the primary contributor, as determined by Gini, permutation importance (PFI), and the LIME algorithm.ConclusionsOur methods not only effectively predicted extended PLOS but also identified risk factors that can be utilized for future treatments. The XGBoost model developed in this study is easily accessible through the deployed web application and can aid in clinical research.

Age is a significant risk factor for mild cognitive impairment (MCI) and Alzheimer’s disease (AD) and identifying brain age patterns is critical for comprehending the normal aging and MCI/AD processes. Prior studies have widely established the univariate relationships between brain regions and age, while multivariate associations remain largely unexplored. Herein, various artificial intelligence (AI) models were used to perform brain age prediction using an MRI dataset (n = 825). The optimal AI model was then integrated with the feature importance methods, namely Shapley additive explanations (SHAP), local interpretable model-agnostic explanations, and layer-wise relevance propagation, to identify the significant multivariate brain regions hierarchically involved in this prediction. Our results showed that the deep learning model (referred to as AgeNet) outperformed conventional machine learning models for brain age prediction, and that AgeNet integrated with SHAP (referred to as AgeNet-SHAP) identified all ground-truth perturbed regions as key predictors of brain age in semi-simulation, demonstrating the validity of our methodology. In the experimental dataset, when compared to cognitively normal (CN) participants, MCI exhibited moderate differences in brain regions, whereas AD showed highly robust and widely distributed regional differences. Individualized AgeNet-SHAP regional features further showed associations with clinical severity scores in the AD continuum. These results collectively facilitate data-driven explainable AI approaches for disease progression, diagnostics, prognostics, and personalized medicine efforts.

For predictive analysis and automatic classification, Deep Neural Networks (DNNs) are investigated and visualized. All the DNNs used for Automatic Target Recognition (ATR) have inbuilt feature extraction and classification abilities, but the inner working gets more opaque rendering them a black box as the networks get deeper and more complex. The main goal of this paper is to get a glimpse of what the network perceives in order to classify Moving and Stationary Target Acquisition and Recognition (MSTAR) targets. However, past works have shown that classification of targets was performed solely based on clutter within the MSTAR data. Here we show that the DNN trained on the MSTAR dataset classifies only based on target information and the clutter plays no role in it. To demonstrate this, heatmaps are generated using the Gradient-weighted Class Activation Mapping (Grad-CAM) method to highlight the areas of attention in each input Synthetic Aperture Radar (SAR) image. To further probe into the interpretability of classifiers, reliable post hoc explanation techniques are used such as Local Interpretable Model-Agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP) to approximate the behaviour of a black box by extracting relationships between feature value and prediction.

Analysis of electrocardiograms (ECG) is vital to identify cardiological problems; however, traditional interpretation requires a lot of clinical knowledge and is influenced by inter-observer variability. In order to solve these problems, this research presents an interpretable deep learning framework that fuses Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (BiLSTM) networks, and an Attention mechanism for multi-label ECG classification. In addition to employing the PTB-XL dataset, the system commits to clinical trust by integrating explainability methods SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-Agnostic Explanations) to enhance transparency.The model reaches a macro F1-score of 0.369 and an ROC-AUC of 0.919, reflecting its high capability of simultaneous multi-class identification of different cardiac diseases. The SHAP Attention superposition helps in providing easy-to-understand explanations, which enable the doctor to see the time-related and shape-related factors of the ECG waveform.This paper focuses on the importance of combining accuracy with explanation in AI healthcare systems and provides a robust, transparent method for computer-assisted cardiac diagnosis.

Introduction: The connection between green innovation and firm performance on the industrial sector of Pakistan is explored by this study, which implements the versatile explainable artificial intelligence (XAI) approaches to reveal the knowledge path with transparent explanations of complex decisions. Objective: It adopts a comparative approach to SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) model in explaining green innovation endeavors into an elucidation of practical business gains guaranteed. Method: On a large scale, study access a comprehensive dataset of 284 manufacturing companies of diverse industries within Pakistan and develop a machine learning model to forecast the firm indicators of performance whilst being interpretable using XAI methods. Results: Research has found that green innovation has a positive impact that is statistically significant on firm performance and the drivers which have been identified are described as environmental compliance, resource efficiency and sustainable product development. The above comparative analysis of SHAP and LIME shows that both methods, in their turn, yield good insights, but SHAP shows more consistent global explanations whereas LIME is better at explaining individual instances. Conclusions: By having the contribution in the growing literature on sustainable business practices in emerging markets, this research is practical in giving the right directions to policymakers and business leaders in Pakistan. Suggestions are given to the policy makers and the firms in Pakistan indicating that strategic investments need to be made in process innovation and regulatory frameworks. This study highlights the promise of XAI to help in de-mystifying the complex nature of relationships in sustainability-related performance.

Significant advancements have been achieved recently in the use of artificial intelligence (AI) to solve healthcare-related problems. A potential method for identifying and treating many illnesses is machine learning (ML). The “Breast Cancer Dataset” was used in this work to perform classification tasks using ML techniques including Support Vector Machine (SVM), Random Forest (RF) and MultiLayer Perceptron (MLP) for identifying the forms of breast cancer, either Benign or Malignant. The SVM, RF, and MLP achieved impressive PR scores exceeding 0.98. However, MLP exhibited superior performance among these algorithms, achieving an exceptional AUC value of 99.71. Conversely, SVM slightly outperformed the other algorithms in terms of training accuracy, with a value of 97.90%. This work used shapley additive explanations (SHAP) and local interpretable model-agnostic explanations (LIME) to shed light on how these categorization algorithms made judgments. These interpretability methodologies showed that the majority of characteristics consistently predicted the presence of malignant or benign cancer types. This study demonstrates how ML algorithms' decision-making procedures may be understood, possibly promoting the use of ML in clinical settings for medical diagnosis.

Explainable artificial intelligence for predicting dengue outbreaks in Bangladesh using eco-climatic triggers.

The conventional criteria of Intrusion Detection Systems (IDS) need to evolve because they fail to detect modern cyber security threats adequately. The advantages of machine learning (ML) and deep learning (DL) models in IDS functionality are limited by the inability to provide explanations which prevents cybersecurity professionals from validating decisions. The research analyzes DL-based IDS performance and interpretability standards through the examination of the NSL-KDD dataset. A screening process identified models that combined high reliability and accuracy numbers as selection candidates. feedforward neural networks (FNN), convolutional neural networks (CNN), and recurrent neural networks (RNN)—findings revealed that CNN achieved the greatest accuracy of 94.2% along with an AUC of 0.97 exempting FNN (91.3%) and RNN (93.8%). The effective extraction of spatial features from network traffic data by CNN models leads to its higher performance. The "black-box" nature of CNNs within DL models makes them difficult to understand because they remain concealed from users. The research integrated local interpretable model-agnostic explanations (LIME) and Shapley additive explanations (SHAP) to interpret decisions at the feature level. The implemented methods did not result in substantial accuracy improvements yet they made classification decisions more trustworthy and understandable by indicating the important features involved. Furthermore, in these developments there exist technical obstacles associated with high processing expenses and performance versus deployment speed balancing requirements. Research initiatives should focus on developing explainability techniques that maintain high-performance rates and excellent interpretability in IDS systems.

Machine learning models are increasingly used for streamflow prediction due to their promising performance. However, their data-driven nature makes interpretation challenging. This study explores the interpretability of a Random Forest model trained on high streamflow events from a hydrological perspective, comparing methods for assessing feature influence. The results show that the mean decrease accuracy, mean decrease impurity, Shapley additive explanations, and Tornado methods identify similar key features, though Tornado presents the most notable discrepancies. Despite the model being trained with events of considerable temporal variability, the last observed streamflow is the most relevant feature accounting for over 20% of importance. Moreover, the results suggest that the model identifies a catchment region with a runoff that significantly affects the outlet flow. Accumulated local effects and partial dependence plots may represent first infiltration losses and soil saturation before precipitation sharply impacts streamflow. However, only accumulated local effects depict the influence of the scarce highest accumulated precipitation on the streamflow. Shapley additive explanations are simpler to apply than the local interpretable model-agnostic explanations, which require a tuning process, though both offer similar insights. They show that short-period accumulated precipitation is crucial during the steep rising limb of the hydrograph, reaching 72% of importance on average among the top features. As the peak approaches, previous streamflow values become the most influential feature, continuing into the falling limb. When the hydrograph goes down, the model confers a moderate influence on the accumulated precipitation of several hours back of distant regions, suggesting that the runoff from these areas is arriving. Machine learning models may interpret the catchment system reasonably and provide useful insights about hydrological characteristics.

Lungs and kidneys are two vital and frequently injured organs among critically ill patients. In this study, we attempt to develop a weaning prediction model for patients with both respiratory and renal failure using an explainable machine learning (XML) approach. We used the eICU collaborative research database, which contained data from 335 ICUs across the United States. Four ML models, including XGBoost, GBM, AdaBoost, and RF, were used, with weaning prediction and feature windows, both at 48 h. The model’s explanations were presented at the domain, feature, and individual levels by leveraging various techniques, including cumulative feature importance, the partial dependence plot (PDP), the Shapley additive explanations (SHAP) plot, and local explanation with the local interpretable model-agnostic explanations (LIME). We enrolled 1789 critically ill ventilated patients requiring hemodialysis, and 42.8% (765/1789) of them were weaned successfully from mechanical ventilation. The accuracies in XGBoost and GBM were better than those in the other models. The discriminative characteristics of six key features used to predict weaning were demonstrated through the application of the SHAP and PDP plots. By utilizing LIME, we were able to provide an explanation of the predicted probabilities and the associated reasoning for successful weaning on an individual level. In conclusion, we used an XML approach to establish a weaning prediction model in critically ill ventilated patients requiring hemodialysis.

Rising shipping emissions greatly affect greenhouse gas (GHG) levels, so precise fuel consumption forecasting is essential to reduce environmental effects. Precision forecasts using machine learning (ML) could offer sophisticated solutions that increase the fuel efficiency and lower emissions. Indeed, five ML techniques, linear regression (LR), decision tree (DT), random forest (RF), XGBoost, and AdaBoost, were used to develop ship fuel consumption models in this study. It was found that, with an R² of 1, zero mean squared error (MSE), and a negligible mean absolute percentage error (MAPE), the DT model suited the training set perfectly, while R² was 0.8657, the MSE was 56.80, and the MAPE was 16.37% for the DT model testing. More importantly, this study provided Taylor diagrams and violin plots that helped in the identification of the best-performing models. Generally, the employed ML approaches efficiently predicted the data; however, they are black-box methods. Hence, explainable machine learning methods like Shapley additive explanations, the DT structure, and local interpretable model-agnostic explanations (LIME) were employed to comprehend the models and perform feature analysis. LIME offered insights, demonstrating that the major variables impacting predictions were distance (≤450.88 nm) and time (40.70 < hr ≤ 58.05). By stressing the most important aspects, LIME can help one to comprehend the models with ease.

In the field of network security, botnet attacks pose a significant challenge, exploiting networks of infected devices to launch sophisticated threats. As these attacks evolve, the need for effective detection methods becomes increasingly critical. This study proposes an explainable Machine Learning (ML) model that aims to identify and categorize botnet attacks, and it investigates the efficiency of different Explainable Artificial Intelligence (XAI) techniques for Intrusion Detection Systems (IDS). To train our models, we employed a subset of the NCC-2 dataset, which includes a mix of normal traffic and seven different types of botnet attacks across three sensors. The ML techniques selected for this research are Random Forest (RF), Extra Trees (ET), Decision Tree (DT), and K-Nearest Neighbors (KNN), with a GridSearch cross-validation approach for optimal hyperparameter tuning. We also explored the effects of class balance through Synthetic Minority Oversampling (SMOTE) and Random Undersampling. The models’ performance was rigorously tested using the Matthews Correlation Coefficient (MCC) and Macro F1-score, with the ET model demonstrating superior results of 99% MCC and 97% F1-score, respectively. To enhance the interpretability of the ET model’s decision-making process, we integrated three XAI techniques: SHapley Additive exPlanations (SHAP), Dalex, and Local Interpretable Model-agnostic Explanations (LIME), and evaluated their efficiency. The LIME and Dalex techniques showed efficient construction times for IDS.

Explainable artificial intelligence-machine learning models to estimate overall scores in tertiary preparatory general science course