The healthcare domain generates huge amounts of complex unstructured data, which includes conversations between patients and healthcare professionals, processing the unstructured nature of clinical descriptions throws challenges for active analysis and decision-making. Natural Language Processing (NLP) has gained significant attention in the healthcare domain for its potential to analyze and forecast outcomes from unstructured medical text data. Text data such as handwritten clinical notes, summary of the discharge, and report of the radiology encompass measurable statistics that can aid in medical research, patient care, and the decision-making process. The intricacy of the unstructured data poses challenges in predicting and analyzing the data. This paper proposes a comprehensive NLP framework with its techniques to analyze and predict specific diseases, such as Birth control, Depression, Diabetes, and High Blood Pressure, based on their relevant keywords built on the medical repository data. Data preprocessing procedures such as Named Entity Recognition (NER), tokenization, and embeddings are applied to extract the common words encountered in unstructured clinical data. Machine learning algorithms such as the Passive Aggressive Model and Naïve Bayes are applied for the analysis of the keywords and to predict the specific disease. Among the Naïve Bayes and PAR algorithms used, the Aggressive Model with TF-IDF gives more accuracy. Integrating NLP into the clinical workflow aims to enhance the potential to revolutionize medical research and decision-making.