There is a wealth of information available about diseases, environments, organized and unstructured data, and people’s awareness of their own health status, making healthcare a promising new topic of study. Smart watches, fitness bands, sensors, and healthcare apps are all examples of how technological progress is improving the healthcare system. The development of these tools marked a significant advancement in medical care. That has resulted in widespread anxiety about the current health situation and its potential implications. Therefore, precise health data analysis and healthcare services constitute a vital component in the healthcare area. Meanwhile, machine learning is a well-known method that has been widely embraced for healthcare data analysis and prediction. The purpose of ML is to make correct diagnoses and quick decisions. The primary focus of this study is on the need of prompt and correct disease diagnosis. Data imputation, accuracy, outliers, and environmental difficulties are chosen as potential problems, and the best possible solution for each of these is sought for here. Optimal feature selection is a key factor in boosting DNN’s performance. Therefore, an artificial bee colony based method is used to extract the relevant features for the stroke dataset. The effectiveness of the DNN approach is measured using a variety of criteria, including accuracy, precision, and recall. All of the features from the stroke dataset, together with the feature weighting technique, are used in the simulations (ABC-FS). It can be seen that the prediction rate of the DNN technique is greatly enhanced by the feature weighting technique. It is also said that the prediction rate of the DNN model is improved by up to 7.5% when relevant features are included.