Towards a Learning Health System: a SOA based platform for data re-use in chronic infectious diseases

Abstract Information and Communication Technology (ICT) tools can efficiently support clinical research by providing means to collect automatically huge amount of data useful for the management of clinical trials conduction. Clinical trials are indispensable tools for Evidence-Based Medicine and represent the most prevalent clinical research activity. Clinical trials cover only a restricted part of the population that respond to particular and strictly controlled requirements, offering a partial view of the overall patients’ status. For instance, it is not feasible to consider patients with comorbidities employing only one kind of clinical trial. Instead, a system that have a comprehensive access to all the clinical data of a patient would have a global view of all the variables involved, reflecting real-world patients’ experience. The Learning Health System is a system with a broader vision, in which data from various sources are assembled, analyzed by various means and then interpreted. The Institute of Medicine (IOM) provides this definition: “In a Learning Health System, progress in science, informatics, and care culture align to generate new knowledge as an ongoing, natural by-product of the care experience, and seamlessly refine and deliver best practices for continuous improvement in health and health care”. The final goal of my project is the realization of a platform inspired by the idea of Learning Health System, which will be able to re-use data of different nature coming from widespread health facilities, providing systematic means to learn from clinicians’ experience to improve both the efficiency and the quality of healthcare delivery. The first approach is the development of a SOA-based architecture to enable data collection from sparse facilities into a single repository, to allow medical institutions to share information without an increase in costs and without the direct involvement of users. Through this architecture, every single institution would potentially be able to participate and contribute to the realization of a Learning Health System, that can be seen as a closed cycle constituted by a sequential process of transforming patient-care data into knowledge and then applying this knowledge to clinical practice. Knowledge, that can be inferred by re-using the collected data to perform multi-site, practice-based clinical trials, could be concretely applied to clinical practice through Clinical Decision Support Systems (CDSS), which are instruments that aim to help physicians in making more informed decisions. With 4 this objective, the platform developed not only supports clinical trials execution, but also enables data sharing with external research databases to participate in wider clinical trials also at a national level without effort. The results of these studies, integrated with existing guidelines, can be seen as the knowledge base of a decision support system. Once designed and developed, the adoption of this system for chronical infective diseases management at a regional level helped in unifying data all over the Ligurian territory and actively monitor the situation of specific diseases (like HIV, HCV and HBV) for which the concept of retention in care assumes great importance. The use of dedicated standards is essential to grant the necessary level of interoperability among the structures involved and to allow future extensions to other fields. A sample scenario was created to support antiretroviral drugs prescription in the Ligurian HIV Network setting. It was thoroughly tested by physicians and its positive impact on clinical care was measured in terms of improvements in patients’ quality of life, prescription appropriateness and therapy adherence. The benefits expected from the employment of the system developed were verified. Student’s T test was used to establish if significant differences were registered between data collected before and after the introduction of the system developed. The results were really acceptable with the minimum p value in the order of 10−5 and the maximum in the order of 10−3. It is reasonable to assess that the improvements registered in the three analysis considered are ascribable to this system introduction and not to other factors, because no significant differences were found in the period before its release. Speed is a focal point in a system that provides decision support and it is highly recognized the importance of velocity optimization. Therefore, timings were monitored to evaluate the responsiveness of the system developed. Extremely acceptable results were obtained, with the waiting times of the order of 10−1 seconds. The importance of the network developed has been widely recognized by the medical staff involved, as it is also assessed by a questionnaire they compiled to evaluate their level of satisfaction.