LACO-GAI: A Laplace Copula Generative AI Network Model for Accurate Stroke Disease Prediction

Abstract

Stroke is major wellbeing condition which creates serious hazard to human health and cause of bereavement and disability universal. Several risk factors contribute to its development, including hypertension, diabetes, tobacco use, raised cholesterol levels, and obesity. Conventional methods have been used to predict stroke, however it frequently face demands in achieving enhanced accuracy as well as minimizing errors. To overcome the demands mentioned above, a new model named the Laplace Copula Generative Artificial Intelligence (LACO-GAI) is introduced to deliver precise stroke predictive outcomes with minimal time and error rate. Initially, patient data samples are gathered from relevant datasets during the data acquisition stage. Input information is first transferred to input level, then forwarded to first hidden layer, where pre-processing takes place. This includes handling missing values through Multiple Imputations by Chained Equations (MICE) technique and removing anomalies or outliers using High Contrast Subspace method to compromise model performance. In second hidden level, is executed by Laplace Copula Multivariate feature ranking method, which identifies and retains the most relevant features. Subsequently, the third hidden layer handles the stroke prediction phase by employing Logistic Regression as a binary classification model to accurately forecast the presence or absence of stroke based on a selected set of relevant features. A probability is then used to classify the patient as either stroke or not. To optimize model performance during training, error backpropagation is employed for weight adjustment, and hyperparameters are fine-tuned using the Adaptive Moment Estimation optimizer. Experimental assessment of LACO-GAI model is conducted using various assessment parameters. Quantitatively examined outcomes expose LACO-GAI obtains superior accuracy in stroke disease prediction by lesser time consumption compared to DL techniques.