ATMOSPHERE aims at measuring and improving the trustworthiness of application and cloud services. The concept of trustworthiness is conceived in its multiple dimensions, addressing security, privacy, repeatability, stability, robustness, performance, fairness, explanation among others. The specific metrics and measures for trustworthiness are discussed in deliverable D3.1. Although ATMOSPHERE focuses on developing the measures, models and actuators for improving trustworthiness, as well as providing a framework to run them, ATMOSPHERE will also implement a pilot use case to experiment and demonstrate this features. The pilot use case focuses on Rheumatic Heart Disease (RHD), a heart disease caused by poorly treating a set of infectious diseases that cause damages in the heart walls and valves. If detected, the infections can be easily managed, but if they remain undetected, it can cause severe morbidity. The project will use echocardiography images obtained from children in rural areas to automatically evaluate the risk of RHD according to the World Heart Organization parameters. The social impact of such application is very high, as this disease affects children and treatment is affordable and effective while damage if untreated can be definitive and even lead to casualties.
The pilot case involves processing medical data with different level of sensitivity. The project has substituted all identifiable critical data with fake data to avoid facing ethical issues but will use this data fields as samples for verifying the management of privacy and the identification of privacy risks. Therefore, the pilot case is a relevant example for properties such as privacy, reliability, robustness, predictable performance, fairness and explainability. The development and usage of cloud services involve three actors (application developer, application manager and end-user). ATMOSPHERE has identified 31 user stories that describe the interactions of the three types of actors with the platform. These user stories involve actions such as the interaction with the infrastructure, the application delivery, the customisation of the trustworthiness properties, the definition of actuators to correct deviations on the trustworthiness properties, and the dynamic monitoring of applications. These 31 user stories led to identifying 14 requirements to drive the development of the ATMOSPHERE layers. The pilot case is a cornerstone development to demonstrate and guide the functionality and properties of the ATMOSPHERE platform.