The growing awareness of the influence of “what we eat” on lifestyle and health has led to an increase in the use of embedded food analysis and recognition systems. These solutions aim to effectively monitor daily food consumption, and therefore provide dietary recommendations to enable and support lifestyle changes. Mobile applications, due to their high accessibility, are ideal for real-life food recognition, volume estimation and calorific estimation. In this study, we conducted a systematic review based on articles that proposed mobile computer vision-based solutions for food recognition, volume estimation and calorific estimation. In addition, we assessed the extent to which these applications provide explanations to aid the users to understand the related classification and/or predictions. Our results show that 90.9% of applications do not distinguish between food and non-food. Similarly, only one study that proposed a mobile computer vision-based application for dietary intake attempted to provide explanations of features that contribute towards classification. Mobile computer vision-based applications are attracting a lot of interest in healthcare. They have the potential to assist in the management of chronic illnesses such as diabetes, ensuring that patients eat healthily and reducing complications associated with unhealthy food. However, to improve trust, mobile computer vision-based applications in healthcare should provide explanations of how they derive their classifications or volume and calorific estimations.

IntroductionWith the growing prevalence of AI-based systems and the development of specific regulations and standardizations in response, accountability for consequences resulting from the development or use of these technologies becomes increasingly important. However, concrete strategies and approaches of solving related challenges seem to not have been suitably developed for or communicated with AI practitioners.MethodsStudying how risk governance methods can be (re)used to administer AI accountability, we aim at contributing to closing this gap. We chose an exploratory workshop-based methodology to investigate current challenges for accountability and risk management approaches raised by AI practitioners from academia and industry.Results and DiscussionOur interactive study design revealed various insights on which aspects do or do not work for handling risks of AI in practice. From the gathered perspectives, we derived 5 required characteristics for AI risk management methodologies (balance, extendability, representation, transparency and long-term orientation) and determined demands for clarification and action (e.g., for the definition of risk and accountabilities or standardization of risk governance and management) in the effort to move AI accountability from a conceptual stage to industry practice.

No abstract

Although numerous ethical principles and guidelines have been proposed to guide the development of artificial intelligence (AI) systems, it has proven difficult to translate these principles into actionable practices beyond mere adherence to ethical ideas. This is particularly challenging in the context of AI systems for healthcare, which requires balancing the potential benefits of the solution against the risks to patients and the wider community, including minorities and underserved populations. To address this challenge, we propose a shift from one-size-fits-all ethical principles to contextualized case-based ethical frameworks. This study uses an AI-enabled mHealth application as a case study. Our framework is built on existing ethical guidelines and principles, including the AI4People framework, the EU High-Level Expert Group on trustworthy AI, and wider human rights considerations. Additionally, we incorporate relational perspectives to address human value concerns and moral tensions between individual rights and public health. Our approach is based on ”ethics by design,” where ethical principles are integrated throughout the entire AI development pipeline, ensuring that ethical considerations are not an afterthought but implemented from the beginning. For our case study, we identified 7 ethical principles: fairness, agility, precision, safeguarding humanity, respect for others, trust and accountability, and robustness and reproducibility. We believe that the best way to mitigate and address ethical consequences is by implementing ethical principles in the software development processes that developers commonly use. Finally, we provide examples of how our case-based framework can be applied in practice, using examples of AI-driven mobile applications in healthcare.