In the present technical era, healthcare providers generate large amounts of medical data on a day-to-day basis. Produced clinical information is placed away carefully as Electronic Health Record (EHR) as essential information archive of medical clinics. These days Artificial Intelligence (AI) have been developing rapidly in recent years. Particularly, wellbeing data framework can get most advantages from the AI benefits. Specifically, symptom based disease prediction expectation exploration and creation turned out to be progressively mainstream in the medical care area as of late. In the paper, we have proposed a structure to assess the proficiency of applying both Natural Language Processing (NLP) and Machine learning (ML) advances for disease prediction framework. As an example we have interpreted n2c2 heart related disease symptom datasets from DBMI portal. The acquired patient records is in XML format which is parsed and converted to structured format and naive Bayes algorithm is applied for training.