The problem of Arabic texts processing by means of semantic methods becomes more popular in the field of natural language processing because of the specific characteristics of this language. This research features the most extensive method of sentiment analysis of Arabic-text-enriched lexicons, and powerful preprocessing methods. The study is concerned with eradicating shortcomings of stereotyped methods by using a context-driven formula that adequately categorizes sentiments in distinct dialects. Therefore, based on the evaluation provided in this paper, the proposed method of classifying high risk patients will have an 85% success rate and hence is highly reliable. These innovations include the design of particular dictionary containing the specific words of Arabic and the procedure for preprocessing the text providing the standard format for the analysis. As the results of the study indicate, special instruments for Arabic text analysis are required, which provides hope for further development of this problem. The proposed framework can thus be used as a starting point for future work since it has the potential to enhance the natural language processing applications in Arabic language.