AI Standardized Patient Improves Human Conversations in Advanced Cancer Care

Kurtis Haut, Masam Hasan, Thomas Carroll, Ronald Epstein, Taylan Sen, Ehsan Hoque · 2025

Read AI Standardized Patient Improves Human Conversations in Advanced Cancer Care by Kurtis Haut, Masam Hasan, Thomas Carroll, Ronald Epstein, Taylan Sen, Ehsan Hoque (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-02694.txt.

Introduction

Kurtis Haut, Masam Hasan, Thomas Carroll, Ronald Epstein, Taylan Sen, and Ehsan Hoque present a virtual patient for training in hard talks about serious illness. Their paper is from two thousand twenty-five. Clinicians must discuss end-of-life care with honesty and care, yet practice with real actors is costly and hard to schedule.

Their system, SOPHIE, pairs a lifelike on-screen patient with a language model and instant feedback. The feedback is tied to published clinical guidance and is tailored to each user. In a trial run by chance assignment, students and clinicians who trained with it improved in three skills. They showed more empathy, spoke more plainly, and gave patients more say. The tool offers practice on demand, at any hour.

Catalog record

CREATORKurtis Haut, Masam Hasan, Thomas Carroll, Ronald Epstein, Taylan Sen, Ehsan Hoque
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE56 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-02694

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

AI Standardized Patient Improves Human Conversations in Advanced Cancer Care Kurtis Haut, Masam Hasan, Thomas Carroll, Ronald Epstein, Taylan Sen, Ehsan Hoque Abstract Serious illness communication (SIC) in end-of-life care faces challenges such as emotional stress, cultural barriers, and balancing hope with honesty. Despite its importance, one of the few available ways for clinicians to practice SIC is with standardized patients, which is expensive, time-consuming, and inflexible. In this paper, we present SOPHIE, an AI-powered standardized patient simulation and automated feedback system. SOPHIE combines large language models (LLMs), a lifelike virtual avatar, and automated, personalized feedback based on clinical literature to provide remote, on-demand SIC training. In a randomized control study with healthcare students and professionals, SOPHIE users demonstrated significant improvement across three critical SIC domains: Empathize, Be Explicit, and Empower. These results suggest that AI-driven tools can enhance complex interpersonal communication skills, offering scalable, accessible solutions to address a critical gap in clinician education. keywords Advanced Cancer Communication, AI-Training, Serious Illness Communication, Personalized Feedback Systems † † equal-contributors: These authors contributed equally to this work. † † equal-contributors: These authors contributed equally to this work. 1 Introduction Serious illness communication (SIC) in end-of-life care remains one of the most challenging aspects of clinical practice Shilling et al. (2024) ; Pasricha et al. (2020) ; Paladino et al. (2023) ; Myers et al. (2024) . These are high-stakes conversations where clinicians must navigate weighty issues, where a poorly chosen word could have lasting consequences on a patient’s final days and the memories their loved ones carry forward. Low-quality SIC has been associated with poor patient and family prognostic understanding Hagerty et al. (2005) , perceived lack of emotional support Korsch and Negrete (1972) , lower quality healthcare outcomes and higher costs Ha and Longnecker (2010) ; Riedl and Schüßler (2017) ; Stewart (1995) ; Begum (2014) ; for Disease Control et al. (2015) ; Oates et al. (2000) ; Beck et al. (2002) . Communication with advanced-stage cancer patients specifically poses a variety of challenges, including: the volume and complexity of medical information, often fast-paced office visits, and the emotional burden of these life-changing conversations, for clinicians, patients, and their loved ones. Despite their extensive medical training, many physicians struggle to deliver difficult news effectively Back and Curtis (2002) ; Gessesse et al. (2023) ; Bagley (2023) , often resulting in patient anxiety, misaligned treatment decisions, and reduced quality of care Starcke and Brand (2012) ; Arnsten (2009) ; Gruneir et al. (2007) . Also costly is the terms of expensive and potentially burdensome treatments as well as malpractice claims Humphrey et al. (2022) . The American Society of Clinical Oncology and the National Cancer Institute affirm that improving SIC is not only an ethical imperative but also a shared priority across healthcare, benefiting patients, their doctors and the broader medical system Epstein and r.L Street (2007) Gilligan et al. (2017) . The current state-of-the-art for communication training relies on human Standardized Patients (SPs), who provide realistic role portrayals and high-quality, actionable feedback. However, high quality SP-based training is limited by high costs, restricted demographic diversity, and the inherent fatigue and variability that result from repeatedly performing emotionally demanding roles Bosse et al. (2015) ; Gillette et al. (2017) ; Menez et al. (2024) ; Everett et al. (2005) . These limitations make it difficult for clinicians—whose schedules are already constrained—to access on-demand, consistent training. AI-driven systems show promise in overcoming the limitations of traditional standardized patient (SP)-based training by offering automated, scalable, reproducible, and personalized training tailored to a clinician’s schedule. Bowers et al. (2024) provide a comprehensive scoping review of AI-driven virtual patient frameworks designed to teach communication skills to healthcare students Bowers et al. (2024) . Complementarily, Wang et al. (2024) demonstrate that integrating AI with wearable sensors enables a more objective measurement of communication quality in real-patient settings Wang et al. (2024) . Additionally, Kearns et al. (2024) show that human-AI teaming with LLMs can foster more empathetic conversations, as clinicians articulate AI-generated suggestions to support compassionate care Kearns et al. (2024) . Despite promising developments, there is a need to establish whether AI systems can cultivate communication skills that clinicians can autonomously apply in practice. Rather than supporting clinicians with real-time AI prompts during patient encounters, there is a need for asynchronous AI-driven training that enables clinicians to deliver conversations that are effective, empathetic, and empowering—without reliance on AI in real-time clinical conversations. To address this knowledge gap, we formed a collaboration between computer scientists and clinicians specialized in SIC training to create SOPHIE (Standardized Online Patient for Healthcare Interaction Education), a virtual standardized patient platform. SOPHIE is designed to simulate a patient with newly diagnosed, advanced lung cancer who is unaware that her condition is incurable. The avatar displays distress, seeks information, and shows resistance to accepting her prognosis. The case scenario requires clinicians to guide SOPHIE through difficult treatment decisions, ranging from comfort care to life-prolonging chemotherapy that may reduce quality of life. In collaboration with oncologists and palliative care physicians, we developed a highly realistic role requiring key SIC skills Back and Arnold (2009) . The team has extensive experience in developing standardized patient roles for medical education and communication research Elias et al. (2017) ; Epstein et al. (2017) ; Epstein et al. (2001) ; Hoerger et al. (2013) . As a training model, we adopted the MVP (Medical, Value, Planning) communication framework from the Advanced Communication Training (ACT) course from the University of Rochester Medical Center (URMC), which focuses on the 3 Es—Empathize, be Explicit, and Empower Horowitz et al. (2020) . SOPHIE’s clinical scenario challenges users to: 1) assess the patient’s readiness to receive serious news (empower), 2) clearly deliver that news (be explicit), and 3) provide emotional support (empathize). After each interactive, spoken encounter with SOPHIE, clinicians receive immediate, automated, and personalized feedback that emulates the guidance of an expert facilitator. As a foundation to create an accurate cancer patient simulation, we first analyzed oncologist-patient dialogues from validated datasets Sen et al. (2017) ; Epstein et al. (2017) , identifying recurring communication patterns. We then developed a schema-based dialogue manager Kane et al. (2023) powered by large language models (LLMs) that balances conversational flexibility with clinical accuracy. SOPHIE integrates a lifelike virtual avatar, rendered with advanced 3D graphics, capable of engaging in real-time spoken dialogue while displaying appropriate emotional cues (Figure 5 ). The system supports both local deployment and remote access via video conferencing platforms, ensuring accessibility across diverse clinical and educational settings. To provide structured feedback, we established quantitative metrics for three core SIC domains: Empowerment, Empathy, and Explicitness. This combination of realism and automated feedback enables healthcare professionals to repeatedly practice and refine critical SIC skills in a psychologically safe environment. We fine-tuned the technology through an initial qualitative clinical evaluation Ali et al. (2021) , evaluated the feasibility of the idea via a controlled pilot study Haut et al. (2023) . In this study, we conduct a randomized clinical trial of the AI-driven communication training intervention for health professionals. Figure 1 shows our randomized controlled study design. Participants were assigned to either an intervention group using the SOPHIE virtual patient system or a control

Read the complete text