Revolutionizing Clinical Skills Assessment with AI: Latest Version of Treatment.com AI's "Medical Education Suite" Fully Operational at the University of Minnesota

Table of Contents

1. Introduction

2. Background of MES Development: The Intersection of AI and Medical Education

3. What is an AI-Simulated Patient?

4. Implementation Case at the University of Minnesota

5. Technical Features of the Platform

6. Impact on Reducing Educational and Operational Costs

7. Enhancing Objectivity and Learning Effect in OSCE Evaluation

8. International Expansion and Implications for Japan

9. Challenges and Future Prospects

10. Conclusion

---

1. Introduction

The OSCE, which quantitatively measures the clinical skills of medical students, has traditionally relied heavily on human resources such as "standardized patients (SP)" and faculty. As a result, variability in evaluation, preparation burden, and rising costs have become chronic issues. MES is an attempt to fundamentally reassess these issues by automating clinical scenario generation, interaction, and scoring using AI technology.

2. Background of MES Development: The Intersection of AI and Medical Education

Since the COVID-19 pandemic, the demand for remote education and e-learning has increased, leading to a surge in simulation research using LLMs (Large Language Models). For example, Stanford University and others have reported the effectiveness of interactive training with AI patients in their trial operations of "MedSimAI" and "Synthetic Patients." 

Treatment.com AI, considering these trends, has developed a SaaS suite that educational institutions can use immediately, centered around a GLM incorporating over 10,000 clinical reviews.

3. What is an AI-Simulated Patient?

AI patients combine case data with natural language generation models to recreate in real-time the processes of medical history taking, physical examination, test result presentation, and emotional expression. In MES,

• Diagnostic Reasoning: Symptoms change in response to student questions

• Real-Time Scoring: Automatic mapping to evaluation criteria for each learning objective

• Immediate Feedback: Providing individual explanations and relearning resources

  are among the features incorporated.

4. Case Study at the University of Minnesota

At the University of Minnesota Medical School, a high-risk OSCE consisting of 11 stations was conducted for 240 third-year students. There were zero reported technical issues, and administrative costs were reduced by approximately 40% compared to traditional methods. Assistant Dean Professor Claudio Violato commented, "AI dramatically enhanced the reliability and consistency of student evaluations."

5. Technical Features of the Platform

Features	Overview	Educational Benefits
GLM-based AI Patients	Over 700 Case Templates and Instant Customization	Ensuring Diversity Close to Real Clinical Practice
Real-time Scoring	Linked with LCME Standards and Bloom's Taxonomy	Objectivity and Immediacy of Evaluation
Cloud Delivery	Supports Both On-campus and Remote	Location-independent Exam Environment
Analytical Dashboard	Individual and Cohort Analysis, Weakness Diagnosis	Data-driven Curriculum Improvement

6. Impact of Reducing Educational and Operational Costs

• Preparation Time: Significantly reduces human costs such as SP arrangement, script creation, and scoring sheet preparation

• Faculty Load: Automates scoring and feedback, reallocating teaching time to qualitative guidance

• Financial Effect: Estimated 40% reduction in administrative costs by the university (equivalent to hundreds of thousands of dollars annually)

7. Enhancing Objectivity of OSCE Evaluation and Learning Effectiveness

Since AI patients can present the same conditions to all examinees, evaluation bias is reduced. Furthermore, using natural language processing for verbatim recording enables the quantification of communication skills and empathetic attitudes. The clinical reasoning engine equipped with GLM analyzes incorrect answer patterns and automatically generates guidance leading to individual remediation.

8. International Expansion and Implications for Japan

Treatment.com AI is in discussions with medical schools in North America and the UK, with plans to expand into the Asian market in the future. In Japan, the online implementation of OSCE is scheduled for demonstration in the fiscal year 2028, and the MES case could serve as a reference model. In regional medical education, which struggles with uneven distribution of doctors and a shortage of supervising physicians, remote evaluation using AI patients could be a powerful solution.

9. Challenges and Future Outlook

• Bias Management: Consideration for regional and racial differences in clinical data learned by AI models

• Examinee Authentication: Identity verification technology for remote examinations

• Ethical Verification: Transparency and accountability of AI scoring algorithms

• Domestic Legal Regulations: Review of personal information protection and medical device applicability

To address these challenges, the company plans to advance joint research with third-party organizations and publish peer-reviewed papers.

10. Conclusion

The Medical Education Suite is a platform that simultaneously addresses long-standing issues such as "constraints on human resources," "variability in evaluation," and "increasing costs" by supporting clinical skills evaluation with AI. The results of its implementation at the University of Minnesota demonstrated that AI patients can complement SPs, enhancing educational effectiveness and operational efficiency as a practical solution. As CBT and OSCE reforms progress in Japanese medical education, the insights from MES will likely serve as a compass for educational DX.

---

Reference Article

• Treatment.com AI Announcement Article (Aktiencheck.de)

• GlobeNewswire Press Release

  Hicke Y. et al. “MedSimAI: Simulation and Formative Feedback Generation to Enhance Deliberate Practice in Medical Education.” arXiv:2503.05793 (2025).

• Chu S. N. & Goodell A. J. “Synthetic Patients: Simulating Difficult Conversations with Multimodal Generative AI for Medical Education.” arXiv:2405.19941 (2024).