Gestational diabetes mellitus (GDM) is a familiar pregnancy-related complicating factor that represents a certain degree of glucose intolerance at the time of onset or first recognition. We went through many Papers and understood the methodologies used by various authors, their requirements, and the challenges faced by them while doing their respective models. We came to know that, in recent years, several studies on the early identification of GDM have been conducted. GDM is usually detected around 22 and 26 weeks of pregnancy and can put both the mother and the baby at risk. In any case, early estimation is preferable as it may lower the risk. Machine Learning (ML) techniques have been proven to be more impactful than statistical models in terms of prediction. In this respect, the present study presents a set of GDM predictions and recommendations that are machine learning based. The proposed model has three layers: Internet – of – things, fog, and cloud computing. First, the input medical data is read using various sensors such as a SpO2 sensor, a temperature sensor, a pulse sensor, and an ECG. In addition, three ML models are used for classification: Random Forest (RF), Convolution Neural Networks (CNN), and XgBoost (XGB). The results from this model will be visualized using AWS Managed Grafana with the help of Amazon Athena. Around 215 papers have been collected and up to 36 papers have been filtered among them. Each of them was filtered based on various factors. Some papers were excluded based on title, few were excluded based on abstract and conclusions.