The world has been facing the challenge of increased cardiovascular death rate for decades now. The utilization of machine learning techniques coupled with the power of Internet of Things can be an effective solution to this real world problem. This research presents Internet of Things based framework for smart healthcare using hybrid machine learning for monitoring heart disease, a system specifically designed for wearable devices that facilitates heart diseases monitoring. The system uses wearable sensors to collect observable vital signs which is contextualized with data from clinical database for a high efficiency in analysis and prediction. Internet of Things and Machine Learning have gained wide applicability in the healthcare sector. The problem of inaccurate Electrocardiogram interpretation and high dependency on individual interpretation, coupled with delayed analysis and diagnosis of patients and lack of accurate prognosis are what this research solves. Machine learning was used accurate analysis and predictions of diseases while the internet of things components was communicating the server for smartness. The framework consists of some components like data collection using wearable sensors, data analysis center, which uses hybridized machine learning algorithms, to quickly identify potential cardiovascular cases from real-time symptom data. This facilitates real-time communication between doctor and patient through wearable technologies. With this, real-time analysis and monitoring of cardiovascular disease would be possible. This system instead of transferring the raw data collected from patients directly to the health care professionals, sends those data to a python platform for analysis on the local device by feeding the hybrid of collected data to Random Forest, Naïve Bayes and Support Vector Machine (SVM) algorithms to analyze and monitor features extracted from clinical databases and wearable sensors and then classify a patient as “Negative” or “Positive” for heart disease. In this research, Object Oriented Methodology and Analysis was used with arduino and python programming languages. The results of our experiment showed that the system was successful in both analysis and classification of a patient’s heart disease with an accuracy of 87.6% for the hybrid machine learning. Random Forest model has a precision and recall that exceeded 100%, 75% precision and 65% recall for Naïve Bayes model and 77% precision and 56% recall for support vector machine. We used UCI clinical dataset of 1025 samples of patients with 25% for test data and 75% for the data training.