This Intranet version is specifically designed and implemented in a special ambulance dedicated to treating burn wounds. It operates independently and does not rely on an internet connection.

It is not connected to a service like OpenAI, but it utilizes the local LLM of MedlibreGPT hosted on premises, which is based on PrivateGPT.

Patient Question

I do not have my prescriptions with me, please check my file.

Answer based on the medical history of the patient

Dr. Michael Truppe Telekonsultation

Implantat Behandlungsvorschlag

link.truppe.at

Draft Research Protocol

AI-Generated Treatment Advice Acceptance in Ambulance Care as Testbed for Remote Treatment Advice for Burn Wounds.

User Interaction Analysis Heatmap and Session Recording

For the Intranet version, we use MATOMO Intranet Analytics.

Microsoft Clarity - Free Heatmaps & Session Recordings

Clarity is a free user behavior analytics tool that helps you understand how users are interacting with your website through session replays and heatmaps.

link.truppe.at

Title

"Evaluation of AI-Generated Treatment Advice for Burn Injuries in Pre-Hospital Settings"

Objective

To assess the acceptance and reliability of AI-generated treatment advice among burn injury patients in ambulances, compared to traditional physician advice.

Hypothesis

More than 50% of patients will favor AI-generated advice over direct physician consultation for relevance, accuracy, and usefulness.

Methodology

• Participants: 150 adults with burn injuries.
• Procedure: Patients receive AI-generated advice on tablets; later compare it with physician advice. Physicians review AI advice for errors. Tablet interactions are monitored.
• Metrics: Patient acceptance surveys, physician accuracy assessments, and interaction data.

Analysis

• Primary Outcome: Patient acceptance rate of AI advice.
• Secondary Outcomes: Quality ratings of AI advice, physician accuracy assessments, interaction data analysis.
• Statistical Tools: Binomial, McNemar's tests, t-tests, regression analysis, and qualitative interviews.

Sample Size

• Required: 150 patients for a significant detection of a 20% difference in acceptance rate.

Ethical Aspects

• Adherence to human subject research ethics and data confidentiality.

Implications

• Insights into AI's role in emergency care and its potential in healthcare.

Conclusion

This study aims to understand patient acceptance of AI in emergency care, potentially revolutionizing pre-hospital patient management.