Evaluation of an AI-based medical application using AI-generated methods: Student experiences with a case study on “patient preference predictors”

Bernd Steffensen; Bettina von Römer

Download Paper | Permalink

Conference: 2025 ASEE Annual Conference & Exposition
Location: Montreal, Quebec, Canada
Publication Date: June 22, 2025
Start Date: June 22, 2025
End Date: August 15, 2025
Conference Session: Engineering Ethics Division (ETHICS) Technical Session - GenAI in ethics education
Tagged Division: Engineering Ethics Division (ETHICS)
Page Count: 18
DOI: 10.18260/1-2--56460
Permanent URL: https://peer.asee.org/56460
Download Count: 5

Paper Authors

biography

Bernd Steffensen University of Applied Sciences Darmstadt/European University of Technology orcid.org/0000-0003-3639-3531

visit author page

Studied Administrative Sciences and Sociology at the Universities in Kiel, Bielefeld (Germany), and Lancaster (UK). Doctorate in Sociology from the University of Bielefeld. Worked from 1992-2000 with Academy for Technology Assessment in Baden-Wuerttemberg

visit author page

biography

Bettina von Römer University of Applied Sciences Darmstadt

visit author page

Bettina von Römer studied sociology at the Universities of Marburg and Bielefeld (Germany). Since 2013 she has been a lecturer at the Darmstadt University of Applied Sciences, Department of Social Sciences. Her research topics are gender studies and the impact of information technology on society.

visit author page

Download Paper | Permalink

Abstract

In the case of serious illnesses, it often happens that difficult treatment-related decisions have to be made that patients themselves can no longer make. In cases where there is no living will, “patient preference predictors” are designed to help. Based on data analysis and supported by AI, these make predictions about the patient's expected will in order to make an appropriate decision. A comprehensive discussion of the ethical aspects and the advantages and disadvantages of such AI-based decision support in healthcare can be found in the literature. As part of a course on the topic of technology assessment, students should evaluate this AI-based IT application and weigh up its advantages and disadvantages. The course is structured into several weekly sessions and a one-day seminar (8 hours). The day seminar serves to simulate a fictitious citizen participation process in which students are confronted with the various considerations and value judgments of experts who comment on the topic of “Patient Preference Predictors”. In order to make the participation of experts from different fields (doctors, ethicists, lawyers, church representatives or patient rights advocates) more vivid and transparent for the students, avatars were created for these people using the Dall-E software. Using the perplexity and ChatGPT software, short statements (3 to 5 minutes) were written that concisely present the fundamental positions of the aforementioned experts. In a further step, avatars and the short statements were merged using D-ID software, so that the avatars were animated and could “present” their statements via video. The basic premise in designing the one-day seminar was that it would not be possible for the students to acquire the relevant knowledge in a short time by reading. The visual representation of the avatars clarifies the different people and makes the diversity of their positions clear. A detailed evaluation with a largely standardized questionnaire and an analysis of qualitative statements shows that this teaching format is well suited to helping students engage with such ethical issues and understand and weigh up the various perspectives. This format enables students to deal with a challenging ethical issue in an exemplary way.

Citation
Format

Steffensen, B., & von Römer, B. (2025, June), Evaluation of an AI-based medical application using AI-generated methods: Student experiences with a case study on “patient preference predictors” Paper presented at 2025 ASEE Annual Conference & Exposition , Montreal, Quebec, Canada . 10.18260/1-2--56460

TY - CPAPER
AB - In the case of serious illnesses, it often happens that difficult treatment-related decisions have to be made that patients themselves can no longer make. In cases where there is no living will, “patient preference predictors” are designed to help. Based on data analysis and supported by AI, these make predictions about the patient&#39;s expected will in order to make an appropriate decision. A comprehensive discussion of the ethical aspects and the advantages and disadvantages of such AI-based decision support in healthcare can be found in the literature.
As part of a course on the topic of technology assessment, students should evaluate this AI-based IT application and weigh up its advantages and disadvantages. The course is structured into several weekly sessions and a one-day seminar (8 hours). The day seminar serves to simulate a fictitious citizen participation process in which students are confronted with the various considerations and value judgments of experts who comment on the topic of “Patient Preference Predictors”.
In order to make the participation of experts from different fields (doctors, ethicists, lawyers, church representatives or patient rights advocates) more vivid and transparent for the students, avatars were created for these people using the Dall-E software. Using the perplexity and ChatGPT software, short statements (3 to 5 minutes) were written that concisely present the fundamental positions of the aforementioned experts. In a further step, avatars and the short statements were merged using D-ID software, so that the avatars were animated and could “present” their statements via video.
The basic premise in designing the one-day seminar was that it would not be possible for the students to acquire the relevant knowledge in a short time by reading. The visual representation of the avatars clarifies the different people and makes the diversity of their positions clear. A detailed evaluation with a largely standardized questionnaire and an analysis of qualitative statements shows that this teaching format is well suited to helping students engage with such ethical issues and understand and weigh up the various perspectives. This format enables students to deal with a challenging ethical issue in an exemplary way.
AU - Bernd Steffensen
AU - Bettina von Römer
CY - Montreal, Quebec, Canada
DA - 2025/06/22
PB - ASEE Conferences
TI - Evaluation of an AI-based medical application using AI-generated methods: Student experiences with a case study on “patient preference predictors”
UR - https://peer.asee.org/56460
DO - 10.18260/1-2--56460
ER -