Heart strokes are a significant global health concern, profoundly affecting the wellbeing of the population. Many research endeavors have focused on developing predictive models for heart strokes using ML and DL techniques. Nevertheless, prior studies have often failed to bridge the gap between complex ML models and their interpretability in clinical contexts, leaving healthcare professionals hesitant to embrace them for critical decision-making. This research introduces a meticulously designed, effective, and easily interpretable approach for heart stroke prediction, empowered by explainable AI techniques. Our contributions include a meticulously designed model, incorporating pivotal techniques such as resampling, data leakage prevention, feature selection, and emphasizing the model’s comprehensibility for healthcare practitioners. This multifaceted approach holds the potential to significantly impact the field of healthcare by offering a reliable and understandable tool for heart stroke prediction. In our research, we harnessed the potential of the Stroke Prediction Dataset, a valuable resource containing 11 distinct attributes. Applying these techniques, including model interpretability measures such as permutation importance and explainability methods like LIME, has achieved impressive results. While permutation importance provides insights into feature importance globally, LIME complements this by offering local and instance-specific explanations. Together, they contribute to a comprehensive understanding of the Artificial Neural Network (ANN) model. The combination of these techniques not only aids in understanding the features that drive overall model performance but also helps in interpreting and validating individual predictions. The ANN model has achieved an outstanding accuracy rate of 95%.

The chronic metabolic disorder known as diabetes mellitus, which is defined by hyperglycemia, poses a significant threat to the health of people all over the world. The categorization is broken down into two primary categories: Type 1 and Type 2, with each category having its own unique causes and approaches to treatment. It is very necessary for the effective management of illnesses to have both the prompt detection and the exact prediction of outcomes. The applications of machine learning and data mining are becoming increasingly important as tools in this setting. The current research study analyses the usage of machine learning models, specifically Voting Ensembles, for the goal of predicting diabetes. Specifically, the researchers were interested in how accurate these models were. Using GridSearchCV, the Voting Ensemble, which consists of LightGBM, XGBoost, and AdaBoost, is fine-tuned to manage outliers. This may be done with or without the Interquartile Range (IQR) pre-processing. The results of a comparative analysis of performance, which is carried out, illustrate the benefits that are linked with outlier management. According to the findings, the Voting Ensemble model, when paired with IQR pre-processing, possesses greater accuracy, precision, and AUC score, which makes it more acceptable for predicting diabetes. Despite this, the strategy that does not use the IQR continues to be a workable and reasonable alternative. The current study emphasizes both the significance of outlier management within the area of healthcare analytics and the effect of data preparation procedures on the accuracy of prediction models. Both of these topics are brought up because of the relevance of the current work.

The function of network intrusion detection systems (NIDS) in protecting networks from cyberattacks is crucial. Many of the more conventional techniques rely on signature-based approaches, which have a hard time distinguishing between various types of assaults. Using stacked FT-Transformer architecture, this research suggests a new way to identify intrusions in networks. When it comes to dealing with complicated tabular data, FT-Transformers—a variant of the Transformer model—have shown outstanding performance. Because of the inherent tabular nature of network traffic data, FT-Transformers are an attractive option for intrusion detection jobs. In this area, our study looks at how FT-Transformers outperform more conventional machine learning (ML) methods. Our working hypothesis is that, in comparison to single-layered ML models, FT-Transformers will achieve better detection accuracy due to their intrinsic capacity to grasp long-range correlations in network traffic data. We also test the FT-Transformer model on several network traffic datasets that include various protocols and attack kinds to see how well it performs and how generalizable it is. The purpose of this research is to shed light on how well and how versatile FT-Transformers perform for detecting intrusions in networks. We aim to prove that FT-Transformers can secure networks from ever-changing cyber threats by comparing their performance to that of classic ML models and by testing their generalizability.