Background and Project Overview

Air rescue teams operate under extreme time pressure and must make critical decisions within seconds. In addition to delivering life-saving treatment, they are required to document the treatment they provide. Currently, this documentation is done either on paper or using digital tools that were not specifically designed for emergency situations. In both cases, the documentation relies on a retrospective reconstruction from memory. This often results in incomplete or error-prone records—an unsatisfactory situation, as even minor documentation errors can have serious consequences for subsequent treatment and patient well-being.

HeliDoc offers a solution, as it is a user‑centered real-time application designed to simplify documentation tasks for air rescue teams. It has already been successfully tested in collaboration with the Christophorus Flugrettungsverein (ÖAMTC) and ARA Flugrettung. Building on these tests, the present project further develops key innovative components: 1) An offline, locally running AI voice assistant for automated documentation support (human-in-the-loop)  during missions, 2) and a highly secure live data stream that provides early information to target hospitals. The main idea is to significantly improve documentation quality, patient safety, and the continuity of information between preclinical and clinical settings.

Aims

The overarching aim of the project is to expand the functionality of the HeliDoc app on scientifically sound basis to enhance the safety, efficiency, and documentation quality of air rescue operations. Specifically, this includes:

• Developing and evaluating a locally operating, AI‑supported voice assistant that helps emergency teams to structure documentation in real time but does make decisions on its own (human‑in‑the‑loop).
• Implementing state‑of‑the‑art cybersecurity standards for an encrypted live data stream for the exchange of information between emergency teams and target hospitals. This enables medical staff to prepare early and appropriately for incoming patients.
• Reducing documentation time, error rates, and cognitive load. As a result, time‑to‑first‑treatment is improved and patient safety is enhanced.

Work Packages and Methods

An offline-capable AI voice assistant based on a locally deployed Small Language Model (SLM) is integrated into the existing HeliDoc architecture. To ensure safe and effective use in a medical context, the development is carried out in close collaboration with AI experts. The model is trained on Austrian medical terminology and adapted to the specific operational conditions of air rescue missions before being integrated into the app. Its core purpose is to enable voice-based completion of documentation fields—always under full user control (human in the loop).

In addition, a secure live data stream is implemented to transmit emergency medical documentation to receiving hospitals. This allows hospital staff to remain continuously informed about the patient’s condition and to take appropriate preparatory measures before patient arrival. To ensure data security, the system relies on state-of-the-art cryptographic methods. All information is encrypted during data transmission and data storage.

Furthermore, the project follows a human-centered design approach, aligning app development and workflows with the needs and expectations of end users.

Benefits

The project helps streamline procedures across the entire emergency care chain:

• Air rescue teams benefit from a significantly reduced documentation burden. AI-supported documentation lowers cognitive load under time pressure and allows teams to focus primarily on patient care.
• Hospitals benefit from early access to structured mission data, improving clinical preparation, resource allocation, and time to first treatment.
• Patients benefit from safer, faster, and better coordinated care throughout the entire rescue chain, leading to improved treatment outcomes and increased survival chances.
• Researchers and system developers gain access to accurate, time-synchronized, and structured preclinical data. These data support further improvements in healthcare quality.
• The project contributes significantly to the digital transformation of emergency medicine, demonstrating how AI, cybersecurity, and human-centered design can improve processes in highly specialized environments.
• Emergency care processes become better coordinated, enabling faster treatment and more cost-efficient use of resources—resulting in long-term benefits for the healthcare system

You want to know more? Please contact Magdalena Druml.

Funding