This paper presents the decision tree computation Diabetes Prediction Algorithm (DPA) for the early measure of diabetes dependent upon the datasets. The datasets are assembled by using Internet of Things (IOT) Diabetes Sensors, contains 15000 records, out of which 11250 records are used for getting ready explanation and 3750 are used for testing reason. The proposed computation DPA yielded a precision of 90.02 %, disposition of 92.60%, and exactness of 89.17% and error speed of 9.98%. Further, the proposed computation is differentiated and existing strategies. By and by there are different computations open which are not completed definite and DPA has an effect. AI is one of the parts of man-made reasoning that permits the improvement of PC frameworks that can gain from encounters without being the need of programming it for each example. AI is essential need of the present condition to take out human exertion just as pretend higher computerization with fewer mistakes. This paper centers on the survey of Early Diabetes discovery utilizing AI procedures and location of the regularly happened messes with it-principally Diabetic retinopathy and diabetic neuropathy. The informational index utilized in a large portion of the concerned writing is Pima Indian Diabetic Data Set. Early diabetes recognition is critical as it assists with lessening the lethal impacts of the diabetes. Different AI strategies like fake neural organization, head part, choice trees, hereditary calculations, Fuzzy rationale and so on have been talked about and analyzed. This paper initially presents the essential thoughts of diabetes and afterward depicts the different strategies used to distinguish it.