Technology improvements have allowed for the collection of massive amounts of data, particularly in the biological and healthcare industries. Data is typically unstructured and can take many forms, including but not limited to images, audio, text, etc. That’s why it’s such a challenge to sift through all the data we’ve collected and find the nuggets that really matter. It’s also been noted that early detection of sickness drastically cuts down on potential human casualties. Regrettably, pertinent data cannot be extracted efficiently to facilitate sound decisions. On the flip side, data mining is a burgeoning discipline that reveals hidden significance in large data sets. Therefore, two methods are proposed to address the aforementioned concerns, namely missing data imputation and outlier detection. These methods are like an upgraded version of CNN and KMPSO. In order to properly compute missing value, the enhanced CNN method is used, and an enhanced weight function is incorporated into the CNN method. However, to deal with the outliers, a hybrid method based on K-Means and PSO is used. A better CNN data imputation method is used, and KMPSO is used to deal with data outliers. The outcomes of the proposed framework are compared to those of different SVM classifiers. After comparing the findings of the suggested framework to those of the other SVM variations, it was found that the latter produced more reliable outcomes.