Your health.

Your guide. ​

In your pocket.

Background
The Healthcare Navigator is an AI-powered digital solution developed to simplify access to medical guidance and help alleviate key challenges in Austria’s healthcare system. These challenges include overwhelmed emergency services and limited health literacy among the general population. By offering low-threshold access to reliable medical information, the tool supports early intervention and informed decision-making.

Overview
In this project, we developed a minimum viable product (MVP) for the Healthcare Navigator, a chatbot that mimics natural human conversation to deliver clear and reliable health guidance. It serves as a standalone application to help users understand symptoms, assess risk, and schedule appropriate appointments. The system is designed to reduce unnecessary emergency visits and direct patients to the most suitable point of care.

The focus was on making the tool intuitive, accessible, and trustworthy. Special attention was given to usability and data privacy to ensure a safe and inclusive experience for all users. A live demonstration will show how the chatbot guides users through a conversation to deliver personalized health advice and actionable next steps.

Key Features
 Symptom Checker and Triage: The chatbot guides users through structured questions to assess symptoms and urgency, offering tailored recommendations.
Appointment Scheduling: If needed, users are connected with appropriate healthcare services to book appointments directly.
Human-like Dialogue: The chatbot is designed to mimic natural conversation, creating a supportive and engaging user experience.
Privacy and Accessibility: The tool is compliant with data protection standards and optimized for different devices and user groups.

Evaluation Study
To assess the quality and suitability of the Healthcare Navigator, a structured evaluation study was conducted focusing on usability, user acceptance, and trust in the AI-assisted system. The evaluation followed a scenario-based usability testing approach combined with post-test questionnaires.

Study Design and Procedure
The study involved 27 adult participants from a heterogeneous background, covering a broad age range and varying levels of digital literacy. Each participant completed predefined task scenarios simulating realistic use cases, including symptom assessment, search for healthcare provider and emergency triage guidance. Testing sessions were conducted in controlled and semi-naturalistic environments using the teams’ hardware.

Following task completion, participants filled out a 17-item questionnaire based on established measurement instruments. All items were normalized to a 5-point Likert scale, with randomized response option ordering to reduce response bias.

Measurement Framework
The evaluation combined three validated constructs:

• System Usability Scale (SUS): Measuring perceived ease of use, learnability, and interaction quality.

• Technology Acceptance Model (TAM): Assessing clarity, perceived usefulness, and intention to use the system.

• Short Trust in Automation Scale (STIAS): Evaluating user confidence, perceived reliability, and trust in the AI assistant.

Key Results
The results demonstrate very high perceived usability, with participants consistently rating the system as easy to use, quick to learn, and self-explanatory. Negative usability attributes such as complexity or inconsistency were largely rejected, indicating no critical usability barriers.

Acceptance results show strong functional approval, particularly regarding clarity of interaction and navigational flow. While long-term usage intention showed more neutral responses, this is interpreted as a reflection of early prototype maturity rather than rejection.

Trust-related findings indicate cautious but appropriate trust in the AI assistant. Participants viewed the system as a supportive decision aid rather than an authoritative medical authority, which aligns well with ethical expectations in healthcare contexts.

Overall Interpretation
Across all dimensions, the Healthcare Navigator was received positively. The absence of strong negative evaluations suggests that the application is well-positioned for further development and real-world testing. The findings support the feasibility of AI-assisted health navigation tools as intuitive, acceptable, and trustworthy complements to existing healthcare services.

Coach: FH-Prof. Andreas Jakl, MSc
Master Program Digital Healthcare
St. Pölten University of Applied Sciences