Text mining is an Artificial Intelligence (AI) technology with Natural Language Processing (NLP) that converts unstructured text and databases into meaningful information. The prediction of disease at earlier stage is an essential and demanding task. Numerous existing methods were introduced for performing efficient disease prediction at an early stage. However, with the inception of new strains the accuracy and time with which the prediction was made was not found to be satisfactory. In order to address these issues, Nonlinear Sammon Projective and Perceptron Boosting (NSP-PB) method for early COVID-19 prediction is introduced. First, a Nonlinear Sammon Projective Pattern Selection is applied to the input patient files to select relevant patterns. Next, Emphasis Perceptron Boosting Classification is applied to the selected relevant pattern to categorize the COVID disease patient files. Here, Emphasis Boost Classifier merges weak learner result to form strong classifier output. This in turn helps to improve COVID disease prediction with higher accuracy and minimal time consumption. Experimental evaluation is carried out for factors such as prediction accuracy, prediction time and error rate with respect to number of patient files. The experimental result reveal that the proposed NSP-PB performs better with a 9% improvement in prediction accuracy, 39% reduction of error rate, and28% faster prediction time for Covid 19 spread prediction compared to existing works.