With the exponential rise in networked infrastructure and advancement of the IoT, the smart city has become an emerging paradigm. The main attributes of a smart city are to monitor the physical world in real time, provide smart services to inhabitants in terms of healthcare, environment, entertainment, transportation, and energy. However, as smart city applications collect wide range of privacy-sensitive information, various issues pertaining to security of data in such systems need to be addressed. On such applications that demands high degree of privacy in an IoT based healthcare setup is security of Electronic Health Records (EHR). Further, as smart city based applications have to react to real time situations, there is a growing demand for more and more computationally efficient algorithms for such applications. In this paper a high capacity, secure and computationally efficient Electronic Health Record (EHR) hiding technique in medical images in an Internet of Things (IoT) driven healthcare system is proposed. The scheme is based on Pixel Repetition Method (PRM) and modular arithmetic. Pixel Repetition Method has been used to scale up the input medical image for cover image generation and modular arithmetic has been used to embed the secret EHR into the scaled-up images. The proposed scheme has been extensively tested for various commonly used medical/test images, and a group of randomly chosen images from UCID repository. Experimental investigations reveal that the proposed system besides being reversible is capable of providing secure and high embedding capacity while maintaining fair imperceptibility. Further the usage of PRM for cover image generation has been found to be highly efficient from computational point of view compared to state-of- art in the area and hence is best suited for the exchange of Electronic Health Records (EHR) in an IoT based healthcare system for smart city applications. •Development of a high capacity and computationally efficient EMR embedding technique.•Lesser computational complexity coupled with high payload and reversibility feature.•Successful development and implementation of computationally efficient PRM technique.•Use of PRM as an effective alternative to Interpolation.•The proposed scheme supports blind extraction.