The objective of this project ‘MedInsight’ is to develop a ‘Personalized Prescription Advisor’ that models the patient reviews and uses Natural Language Processing (NLP) in the evaluation of medicine. The domain of healthcare abounds with many patient reviews that are essential for understanding the effectiveness and side effects of different drugs. However, the scope of this study is much beyond the conventional methodologies such as sentiment analysis or review summarization. Classification of patient reviews based on patient’s condition, resulting in personalized drug recommendations. Insights are drawn from the text of the patient review, with respect to the condition of the patient and their response to different medications, by going through the complete analysis process. The effectiveness of the proposed approach is illustrated with a diligently undertaken case study on real-world patient reviews. The techniques used for this research include Naïve Bayes and Passive Aggressive Classifiers and their bigrams and trigrams. It is a simple and an effective algorithm for probabilistic classification tasks and thus is employed. Further, it is adaptive to changes with the model parameters based on how well the outputs are making confident and correct predictions. Passive Aggressive variants with bigrams and trigrams capture the contextual meaning and dependence from text data, and adding to the classification accuracy.