Adaptable Intensive Care Unit (ICU) system is a key concern for hospitals in developing countries like Bangladesh. Most of the hospital in Bangladesh lack serving proper health service due to unavailability of appropriate, easy and scalable systems. There also appears communication and information gaps between hospital authority and patient?s relative. The aim of this research is to build an adequate system for hospitals to serve the ICU patients with real-time feedback systems. Based on the doctor?s suggestions, we primarily have chosen the main factors for our project. In this paper, we propose a generic architecture, associated terminology and a classificatory model for observing ICU patient monitoring systems with machine learning and cloud computing. Machine learning health prediction is the key concept of this research. IBM Cloud is the platform for this research to store and maintain our data. For our machine learning models, we have chosen the following algorithms: Naive Bayes, Logistic Regression and Decision Tree. For real-time data and information view, we have developed a Mobile Application named ?ICU Patient Management System – IPMS?. Our system architecture is designed in such a way that the machine learning models can train and deploy in a real-time interval by retrieving the data from IBM Cloud and the Cloud information can also be accessed through IPMS in a requested time interval. To help the doctors, the machine learning models will predict the condition of a patient. If the prediction based on the condition gets worse, the IPMS will send an SMS to the duty doctor and nurse for getting immediate attention to the patient. Combining with the cloud storage, distributed database system, machine learning models and mobile application, the project may serve as a complete medical decision for the doctors