User Interface Design for AI-Based Clinical Decision-Support System Preliminary Study Gabriela Beltrão School of Digital technologies Tallinn University Tallinn, Estonia gbeltrao@tlu.ee Iuliia Paramonova School of Digital technologies Tallinn University Tallinn, Estonia juparam@tlu.ee Sonia Sousa School of Digital technologies Tallinn University Tallinn, Estonia scs@tlu.ee Abstract — This paper presents a case study about the initial phases of the interface design for an artificial intelligence-based decision-support system for clinical diagnosis. The study presents challenges and opportunities in implementing a human-centered design (HCD) approach during the early stages of the software development of a complex system. These methods are commonly adopted to ensure that the systems are designed based on users' needs. For this project, they are also used to investigate the users' potential trust issues and ensure the creation of a trustworthy platform. However, the project stage and heterogeneity of the teams can pose obstacles to their implementation. The results of the implementation of HCD methods have shown to be effective and informed the creation of low fidelity prototypes. The outcomes of this process can assist other designers, developers, and researchers in creating trustworthy AI solutions. Keywords - human-centered design; decision support systems; design for trust I. INTRODUCTION Artificial intelligence-based (AI) solutions have gained space in nearly every segment of society for the past decade. Despite the generalization of the term, these systems vary significantly in their goals, functioning and risk posed to the users. For instance, the European Commission proposal for AI regulation [1] categorizes AI technologies according to their risk, from 'minimal' to 'unacceptable'. Based on it, systems that pose a higher risk should be subjected to stricter regulations to ensure safety. Accordingly, from the users' perspective, these systems should also adopt strict and transparent processes in their development to be trusted, a key factor for their adoption and long-term usage [2][3]. The medical sector has seen benefits from using AI-based tools for diagnosis and, in general, as decision-support systems. However, the nature of the activity brings two major challenges: first, from the system development perspective, as it requires large amounts of medical data while following strict privacy regulations and preventing any bias, and the need for expertise from professionals from different fields to establish and assess the quality of the outcomes. Second, apart from being successfully developed, these systems need to be adopted by their target, mainly clinicians and clinical researchers. Such specialized audiences have their concerns, from which two can be highlighted: the effectiveness of the tool and how its adoption will affect their current practice. Trust, in this case, becomes crucial. This study presents the approach adopted for the initial phases of the user interface (UI) design of an AI-based decision- support system for clinical decisions, AI-Mind. The project aims to develop an AI-based system for assessing the risk of developing dementia in patients with mild cognitive impairment (MCI), reducing the time for diagnosis and enabling earlier intervention. Upon completion, it is expected that the system will be able to analyze multimodal data, including electroencephalogram (EEG) input, to assist clinicians and clinical researchers in the diagnosis procedures. Due to the project's complexity, a plural team is involved to ensure that all aspects of development and usage of the tool are covered. Besides the multiplicity of specializations (e.g., neuroscientists, researchers, software developers, machine learning specialists, healthcare management specialists), the outcome needs to fulfill research and industry requirements. While this broad scope is essential for the quality of the final tool, it can pose challenges for the conciliation of internal processes. II. OBJECTIVES The current study unfolds during the early stages of the UI design for AI-Mind. It presents the procedures adopted for the development of UCD methods, namely (1) personas, (2) scenarios of use, and (3) journey maps. These methods are commonly adopted in HCD processes to ensure that the systems are designed based on users' needs; In this project, in addition, they aim to prevent potential trust issues by ensuring that the users' and stakeholders' concerns are adequately addressed. Additionally, they were used guide the development of the prototypes, a crucial phase in the development of the interface. This study aims to present practical suggestions for adopting UCD methods in similar, heterogeneous projects and thus collaborate with other researchers and practitioners. In the face of the challenges posed by the heterogeneity of the teams' views and the uncertainty of the project's early phase, the authors had to explore alternatives to the usual approaches