Long-Term Prefetching for Cloud Medical Imaging Repositories Carlos VIANA-FERREIRA 1 , Sérgio MATOS, Carlos COSTA Instituto de Engenharia Electrónica e Telemática de Aveiro, Universidade de Aveiro Abstract. Healthcare institutions have been outsourcing their IT infrastructure to the cloud. This new paradigm has financial and technological advantages. However, it has also associated some important issues. In the medical imaging field, studies typically involve high volumes of data, leading to increased communication latency in remote access environments. In this context, this paper presents a long-term prefetching solution that dynamically adapts itself to institutional workflows and services. Evaluation tests show that the proposed solution significantly reduces the data access latency. Keywords. Picture Archiving and Communication Systems, Information Storage and Retrieval, Teleradiology, Health Information Management. Introduction Nowadays, with the tremendous amount of ubiquitous computational power, the communications broadband and the Internet resources available as a common utility, medical imaging services are being outsourced to the Cloud [1, 2]. This new paradigm has financial and technological advantages but also potentiates the inter-institutional workflows. However, communication latency still hinders in some cases the adoption of this paradigm for replacement of traditional solution over local area networks. To solve this issue, we have been working on an intelligent cache mechanism [3]. This paper proposes an enhancement to that cache mechanism, an algorithm that fetches medical imaging data to populate the cache, aiming to meet the long-term needs (day, week or month). Usually, solutions of this kind are based on static rules [4] and need to be adjusted whenever there are significant changes in the institution’s work- flow or services provided. As such, the goal of the proposed system is to be able to dynamically adapt over time, according to behavior, workflow and personnel changes. 1. Methods The prefetching algorithm is based on a cycle with a parameterized period (e.g. a day, week or month) with two states: training and prefetching. The training is carried out through the increment of counters. Each counter represents a determined characteristic of the study. A medical imaging object can contain numerous data elements, including 1 Corresponding Author. Digital Healthcare Empowering Europeans R. Cornet et al. (Eds.) © 2015 European Federation for Medical Informatics (EFMI). This article is published online with Open Access by IOS Press and distributed under the terms of the Creative Commons Attribution Non-Commercial License. doi:10.3233/978-1-61499-512-8-1028 1028