ABSTRACT Healthiness is the maximum unique aspect in human existence. However, because of numerous facts like unhealthy diets, bodily inaction, high blood pressure, smoking habits, and many others. Health sicknesses are growing day by day. Among them coronary heart diseases are the essential causes of loss of life. Threat elements like excessive blood strain, excessive blood cholesterol, abnormal pulse rate, diabetes, consuming habits, age being a prime aid for occurring heart illnesses. Due to the lack of accurate scientific assist structures which can be capable enough to find hidden styles of medical records and are expecting sicknesses, humans not able to understand the prevalence of diseases in advance. Every year we lose approximately millions of people due to myocardial infarction (MI) in India. This makes it of utmost importance to predict not only MI but early MI as well. Looking at this huge number guided the right path to work on. At the initial stage of research, the real data is collected from hospitals. This dataset is not sufficient to give to the model. Providing limited information restricts the learning of the model leading to compromised results. To overcome the issue, a new path was to be taken by feeding synthetic data in order to provide information in bulk to the model. Generation of synthetic data required a lot of research from the previous research, internet and expertise. After all the discussions and study, range for different parameters were noted for early MI, MI and NON-MI which formed a base to produce synthetic datasets. Machine leaning model works better on large number of datasets. Dataset is prepared under the guidance of medical doctors and a huge study is done on research paper. There is an urgent need of preparing datasets on MI. As the datasets available is very old datasets and it is not an Indian also. And lots of death in India happened due to heart attack. The information available on media is not a recent one about the patients. In this research the datasets of 453 patients are collected from rural area. Based on this dataset, steps are shown to create synthetic datasets. Keywords: Synthetic datasets, Early MI, MI, NON-MI, Actual datasets.