Efficient Lung Cancer Detection Based on Support Scalar Vector Feature Selection with Fuzzy Optimized-Multi Perceptron Neural Network Using Natural Language Processing

Lung cancer is one of the most frequent types of cancer worldwide, and it has a high fatality rate. To improve survival chances and ensure a successful treatment, early identification of lung cancer is essential. The rapidly developing artificial intelligence discipline of Natural Language Processing (NLP) has great promise for transforming the understanding, identification, and management of lung cancer. Previously, NLP-based lung cancer prediction models may not always produce accurate results, as they rely on complex algorithms to analyze text data. Errors in language processing or interpretation can lead to incorrect predictions and potentially harmful decisions. This study suggests employing Fuzzy Optimized -Multi Perceptron Neural Network (FOMPNN) for lung cancer detection based on NLP and the Relative Support Scalar Vector Feature Selection (RSSCVFS) method for optimal attribute selection. To begin with, preprocessing involves finding null and missing values, and textual summaries based on stemming and tokenization. Then, the Semantic Lexicon-Based Approach (SLBA) is used to find the sentiment score. Furthermore, the sentiment score of each word in the medical summary can be added up to get the overall sentiment of the text. Based on dictionary terms, feature outcomes are added to patient records. Similar generalization patterns based on Linguistic Subset Pattern Mining (LSPM) are used to produce corresponding feature patterns. The RSSCVFS method is used to pick features based on their significance to minimize the feature dimension of the illness effect rate. To classify the risk of lung cancer, the FOMPNN is finally utilized to test the feature's weight. The proposed method proficiently produces better performance than other methods.