This SUSS BME363 End-of-Course Assessment uses one real-world source, a Stanford paper on how an academic medical center adopted large language models through its ChatEHR platform. Across six marked parts you explain where LLMs fall short in medicine, how ChatEHR was built, evaluated and governed, what its automations do, how much value the pilots showed and what slowed implementation.

Question 1

Please refer to the Stanford article titled “Adoption and Use of LLMs at an Academic Medical Center” at https://arxiv.org/abs/2602.00074 or any relevant online sources (with appropriate citations) in order to answer the questions below.

Question 1a

Demonstrate your understanding of the current limitations of large language models (LLMs) in medicine, and the requirements of their safe and effective use in clinical care. (15 marks)

Question 1b

Examine and discuss the capabilities and evaluation framework of the ChatEHR platform. (20 marks)

Question 1c

Describe the concept of automations in the context of the ChatEHR platform. Illustrate your answer with ONE (1) example of how automations can be used in a healthcare setting. (10 marks)

Question 1d

Discuss the governance and team design for the ChatEHR platform. (15 marks)

Question 1e

Analyse the value assessment results of the automations and user interface pilot studies of ChatEHR platform. In your answer, compare the benefits demonstrated by each pilot and explain the factors that influence their estimated value. (20 marks)

Question 1f

Examine the challenges and limitations faced in the implementation of the ChatEHR platform. (20 marks)

Submission Requirements

• Use the official answer template in Appendix A (course code, ECA title, PI number, name, tutorial group, date of submission).
• Complete the declaration on originality, no resubmission for another course, and use or non-use of AI tools under the Academic Integrity Rules.
• If AI tools were used, include the AI Attribution Table (Table A) with the tool, purpose, exact prompts, generated output and how it was used, and keep your original work as evidence.
• Reference all sources, including AI-generated content, in APA style. Submit through Canvas; late work follows the ECA marks deduction scheme.