In a software development project, the usefulness of a system specification depends on the completeness of the requirements. The software requirement is partitioned into two functional and non-functional requirements. The functional requirement is a requirement software must perform or do, whereas the non-functional requirements are the quality attributes which software must-have. Often a software engineer gives a low priority to non-functional requirements which could lead to a failure of the software or it costs a huge amount of money to fill the incomplete non-functional requirements. Although identifying all requirements is difficult, especially when requirements interact with an unpredictable environment. The main cause for the failure of the system is incomplete requirements. Even if identifying the requirement completeness is a challenging task, researchers try to measure the completeness of the system requirement. It is because of not considering whether or not to fulfill all relevant requirements? The non-functional requirements are selected in the domain of health information systems, the attributes which the study used are availability, privacy, performance, security, reliability and usability. Finally, measuring the completeness of requirements is a researchable area due to its difficulty to know the complete requirements. Therefore, in this research, we proposed machine learning techniques for measuring the completeness of non-functional requirements from SRS documents. After a comparative experimental evaluation of - machine learning classification algorithms (SVM, DTand KNN) ,out of which SVM perform best with F1 score of 92% to deterimine wheter the given health requirement documents is complete or not complete .As we recommend that for future work we compare and contrast our result with that of recent deep learning based classification.