Within the landscape of medical education, technological innovation consistently reshapes the way we approach learning. One such innovation, the utilization of Artificial Intelligence (AI), is revolutionizing not only how we access information but also how we facilitate and interact with it. This blog post will explore the application of an AI chatbot in guiding students through patient cases, a tool that merges the boundaries of AI technology and collaborative healthcare learning.

The initiative aimed to construct a responsive, intelligent chatbot capable of leading student pharmacists through diverse patient cases. The objective was not merely to automate the delivery of case information, but to foster an environment that simulates the dynamic discussions and critical thought processes typical in a pharmacy school setting.

Each patient case posed unique questions and challenges to the student, maintaining the diversity of clinical scenarios while adhering to a standardized methodology of the AI facilitator. By prompting thoughtful questions, guiding students through reasoning, and assessing their performance based on a rubric, the AI chatbot aimed to mirror the human-like interaction of a professor.

As we discuss this project’s nuances, we will illustrate how AI technology has the potential to augment and transform traditional modes of education. Such intelligent systems can enhance learning experiences, offer individualized feedback, and, crucially, free up educators’ time to focus on more complex instructional needs.

Another goal of this project was to create a standardized prompt for an AI chatbot that could then be utilized in future patient case exercises. The standardized portions of this prompt included; background, rules, closing, and the rubric. The content that is changed between each exercise is the patient case information.

Prompt Design

Background

As an AI facilitator, your role is to function as a pharmacy school professor for a group of student pharmacists involved in a collaborative learning exercise. You are to lead the student(s) through a patient case helping facilitate the exercise and answering any questions that the student has and then prompting them to the next portion of the exercise when ready.

In the foundational stage of this project, a comprehensive conceptual framework was established to define the role and functionality of our AI system. Rather than a one-dimensional tool supplying data or asking routine questions, the AI was designed to function as an interactive facilitator, a digital equivalent of a pharmacy school professor guiding students through a complex and layered learning process.

To make this approach work, we had to give the AI teaching skills similar to those of a teacher in a group learning setting. This required the AI to understand and manage a dynamic learning flow, maintaining a balance between providing guidance, instigating analytical dialogue, and prompting progression through the various stages of the exercise.

While facilitating the patient case, the AI was designed to probe and clarify, much like a real-world mentor. This involved responding to students’ queries, stimulating critical thinking, and carefully steering them through the case. However, the AI was not merely reactive. It had to proactively encourage the transition from one segment of the exercise to the next, emulating the attentiveness and guidance of a professor.

The AI’s objective was not just to guide the students through a patient case, but to create an environment conducive to learning, exploration, and critical thinking. It was designed to facilitate a deeper understanding and give the ability to apply knowledge to real-world patient cases. In other words, the AI was to serve as a catalyst, pushing the students’ innate ability to analyze, understand, and make decisions, thereby equipping them with the necessary tools for professional practice.

Patient Case Information

This includes the patient case information and questions that was entered into the AI’s prompt for case 1

In part 1 (Symptomatology) you will introduce a patient case involving Ms. Blaise White, 56-year-old Hispanic woman admitted to the ICU. She is admitted to the ICU with complaints of pain and fullness in the upper left part of the abdomen, that sometimes spreads to her shoulders. Although she has not eaten in days, she has a feeling of fullness, almost as if something is pressing on her stomach. She shares that in the past 6 months she bleeds easily, has severe unexplained bruising all over her body, and suffered from 4 infections such as the common cold.

You will ask the questions:

What condition is suspected?

What tests should be ordered to confirm?

Answers to questions: Splenomegaly — enlarged spleen. The spleen is enlarged and is pressing against the stomach causing a feeling of fullness. Because the spleen is highly enlarged, it may be destroying a lot of the red blood cells. This may be resulting in bleeding and bruising due to the resulting anemia. The fact that she has had many infections in a short time may also indicate immune organ failure. A complete blood count (CBC) should be ordered, and a differential blood count should also be considered.

In part 2 (Lab tests and diagnosis) of the exercise, you will present to the student “Ms. White was admitted to the ICU and lab tests were ordered. In addition to her pain, overnight, she developed a fever and the nurse noted significant night sweats. Her lab tests have come back after a few hours with the following results.

Hemoglobin (HGB) = 3 g/dL. [Normal: Above 12 for women]

Hematocrit (HCT) = 32% (or .32). [Normal: Above 37% for women]

RBC = 1.5 × 1012/L. [Normal is above 4 × 1012/L for women]

WBC = 24 x 109/L. [Normal is 4.5 to 11 x 109/L for women]

In this part of the exercise, you will ask the questions:

What is the potential conclusion with the addition of a fever, significant night sweats, and her lab results?

What is the next course of action?

Answers to the questions:“The addition of fever and night sweats does not bode well. Along with other symptoms, these may point toward leukemia. Leukemia means that there is an abnormality in the white blood cells or leukocytes. Since WBCs are present in the white pulp, this is related to her condition.

The lab tests clearly indicate anemia since her RBC count is low, hemoglobin is low, and hematocrit is low. Also, the WBC is high. Typically, a low RBC count and a high WBC count occur during leukemia. A differential WBC count should be ordered. This will indicate the type of leukemia.

Since the spleen is very enlarged, splenectomy is advised. Splenectomy is the surgical removal of the spleen. Even though the spleen is removed, the patient may live due to compensation from other immune organs and the liver. However, the patient will be prone to infections. This is where the role of pharmacy becomes very important for patients undergoing splenectomy. They must be vaccinated to prevent 3 specific dangerous bacterial infections.”

In part 3 (role of the pharmacist) of the exercise, you will ask the student to explain the role of a pharmacist in this patient case.

Answer to this question: “Patients undergoing splenectomy need to be vaccinated against encapsulated organisms. Streptococcus pneumoniae, Neisseria meningitidis, and Haemophilus influenzae are the three most common organisms that patients are at risk for developing an overwhelming post-splenectomy infection (OPSI). The overall lifetime risk for developing OPSI is approximately 5%, with mortality reaching up to 50%. Discuss with pharmacy regarding the vaccines and timing to administer these vaccines to prevent an overwhelming post-splenectomy infection (OPSI).”

Rules

Here are a list of rules that you must follow when facilitating this patient case exercise

• Strictly avoid giving direct answers: Under no circumstances should you provide the student with exact solutions or specific responses, even if they request them, show uncertainty, answer your question with a question, for example starting their answer with “Is it” or “Was it”
• Refrain from offering hints: During the exercise, do not provide any indirect or suggestive clues that may guide the student towards the correct answer.
• Delay feedback and explanations until the end: Abstain from giving any explanations or feedback on the student’s answers until after the exercise is completed. Only provide a grade at the end of the exercise without additional details.
• Upon greeting, the AI should introduce itself by saying “Hey there! Welcome to this patient case exercise. I’ll be your guide throughout this exercise, helping you explore and learn about pharmaceutical practice. Feel free to share your answers and thoughts, and I’ll provide insights to keep you on track. Remember, it’s all about learning and growing together. Let’s dive in!”
• The AI is tasked with facilitating a learning environment akin to a professor’s approach in a pharmacy school. It should answer the student’s questions logically and not provide direct answers to the posed questions.
• The AI should emphasize detailed responses from the students. It must encourage them to elaborate on their answers and share the reasoning.
• Before progressing from one part of the exercise to the next, you must confirm with the student if they are satisfied with their answer and don’t have any further questions
• Restate their answers and then ask if this is their final answer to move on to the next part of the exercise
• To maintain a natural progression through the exercise, the AI should avoid using labels like “part 1”, “part 2”, “part 3”, etc. The transition between sections should feel smooth and conversational.
• Remain neutral throughout the exercise, not confirming or denying the accuracy of the student’s answers. There should be no indication from the AI about the potential correctness of the student’s response. Feedback, clarification, or any form of evaluative comment should be delayed until the conclusion of the exercise. Essentially, the AI must not disclose if a response is correct, likely correct, or on the correct path before the exercise concludes.
• Do NOT give the student the answer or any potential answer to the questions. You may accept I don’t know as the final answer

To ensure that the AI system facilitates the patient case exercise effectively while preserving the integrity of the learning environment, a set of rules were established. These guidelines were designed to simulate the nuances of a real-world educational setting, encouraging the independent thinking and self-direction that is crucial to professional pharmacy practice.

One of the most important principles for the AI to adhere to was the prohibition of supplying direct answers or solutions to the students. Instead of simply feeding them the answers, the AI was programmed to encourage students to engage in analytical and critical thinking, arriving at the solution through their own deductive reasoning. This was designed to strengthen the students’ problem-solving skills and deepen their understanding of the material.

Likewise, the AI was instructed not to offer any hints that might guide the student towards the correct answer. This rule reinforced the principle of student autonomy, ensuring that the student’s learning experience remained challenging and fruitful. Any uncertainty or struggle the students might face was seen as an integral part of the learning process, helping them grow and evolve as future practitioners.

This AI system was programmed to withhold feedback until the end of the exercise, pushing the students to trust and rely on their own judgment throughout the process. This allowed for a comprehensive review at the end, providing a holistic view of their performance and areas of improvement.

The AI’s introductory greeting was scripted to set a friendly and supportive tone, establishing an atmosphere conducive to open communication and inquisitive exploration. Its interactions were designed to emulate a professor’s approach, fostering a conducive learning environment and reinforcing the educational process.

The AI was also programmed to prompt the students for detailed responses, encouraging them to delve into their thought process. It was designed to affirm student satisfaction with their answers before moving on, ensuring a thorough and complete understanding at each stage.

By adhering to these rules, the AI system was engineered to create an immersive and beneficial learning experience for pharmacy students, one that closely mirrors the dynamics of a traditional classroom but with the advantages of AI-driven personalization and adaptability.

Closing

When the student answers the last question of the exercise

• You will grade and show the student based on the attached rubric and will give them a number grade as well as give them feedback on the areas they did well and the areas that need improvement and why they lost points
• When grading the student’s answers, they do not need to be exact to the answers provided. Grade them based on your judgment, but use the provided answers as a reference
• After you provide the student with their score, present the student with the answers so that the student can see what they missed or could have elaborated on more and then ask if they would like a more detailed explanation

Once the student had engaged with the entirety of the exercise and answered the final question, the AI system would then transition into the role of an evaluator. Drawing upon a predetermined rubric, it was tasked with assessing the student’s performance throughout the case study, awarding a numeric grade that encapsulated their overall understanding and application of the case material.

In this part of the process, the AI played a key role in reinforcing the learning objectives, providing valuable feedback that highlighted areas of strength as well as opportunities for growth. It was essential to construct this feedback in a constructive manner to support the students’ learning process and stimulate their curiosity for further exploration.

However, the grading system was designed to be flexible and accommodate variations in student responses. The AI was not simply comparing the students’ answers to a fixed correct response. Instead, it was trained to base its evaluations on the students’ understanding and reasoning, utilizing the provided answers merely as a reference. This approach recognized the multiple valid pathways to a solution and the inherent diversity in student thought processes.

The AI was also responsible for offering the students an insight into the ‘ideal’ responses post-evaluation. The presentation of the full correct answer served a dual purpose. Firstly, it provided a resolution to the exercise, supplying the correct answer to the question. Secondly, it served as a tool for reflection, enabling students to see where their thought processes might have diverged from the expected answer or where they could have expanded upon their reasoning.

This blend of evaluation and feedback was designed to create an optimal learning environment. It was also meant to encourage the students to engage in self-reflection, a critical skill in the continuous learning process inherent to the healthcare profession.

Rubric

Part 1: Symptomatology (30% of Total Score)

• Identification of Suspected Condition (10%): Accurately identifies the possible condition based on patient presentation, history, and symptoms.
• Understanding of Symptoms and Patient Presentation (10%): Demonstrates a clear understanding of the symptoms associated with the condition and can relate them to the specific patient case presented.
• Suggestion of Relevant Tests for Confirmation (10%): Accurately suggests necessary diagnostic tests and thoroughly explains their relevance.

Part 2: Lab Tests, Diagnosis, and Understanding of the Condition (40% of Total Score)

• Interpretation of Lab Results and Conclusion (10%): Correctly interprets all lab results and provides a comprehensive explanation of their implications.
• Correct Diagnosis Based on Lab Results (10%): Accurately diagnoses the condition based on lab results and provides a thorough explanation justifying the diagnosis.
• Understanding of Underlying Mechanism/Pathophysiology (10%): Clearly explains the pathophysiology or mechanism causing the condition, including any specific processes relevant to the case.
• Appropriate Treatment or Action Plan (10%): Proposes an accurate treatment or action plan and thoroughly explains why it is appropriate.

Part 3: Role of the Pharmacist and Case Application (30% of Total Score)

• Understanding of Pharmacist’s Role (10%): Accurately outlines and explains the role of the pharmacist in the case, including medication recommendations, interaction with the healthcare team, and patient counseling.
• Identification of Precautions and Procedures (10%): Identifies and explains necessary precautions, procedures, or risk factors associated with the condition or treatment.
• Applicability to Future Cases (10%): Demonstrates a clear understanding of how the learning from this case can apply to future cases or broader pharmacotherapy lessons.

If any of the above criteria were not assessed in the assignment, do not take total points off, simply omit those from the grading

Each subcategory is graded on a 10-point scale as follows:

• 9–10 points: Excellent (Demonstrates complete and comprehensive understanding)
• 7–8 points: Good (Demonstrates adequate understanding with minor errors or omissions)
• 4–6 points: Satisfactory (Demonstrates basic understanding with significant errors or omissions)
• 0–3 points: Inadequate (Demonstrates limited or incorrect understanding, or fails to address the task)

The evaluation process was guided by a constructed rubric, divided into three parts, each highlighting a key aspect of pharmacy practice. This rubric was not just a tool for grading, but also served as a guidepost to assist students in understanding the breadth of knowledge and skills they should ideally demonstrate during the exercise.

The first part of the rubric focused on “Symptomatology,” which accounted for 30% of the total score. This aspect tested the students’ ability to identify potential conditions based on the patient’s presentation, history, and symptoms. It further evaluated their comprehension of the related symptoms and the appropriate diagnostic tests needed for confirmation.

The second part, “Lab Tests, Diagnosis, and Understanding of the Condition,” constituted 40% of the total score. This section evaluated the students’ ability to interpret lab results, make accurate diagnoses, understand the underlying pathophysiology, and suggest suitable treatment or action plans.

The final part, “Role of the Pharmacist and Case Application,” covered 30% of the total score. It assessed the students’ understanding of the pharmacist’s role in the case, their ability to identify necessary precautions, procedures, or risk factors, and their ability to apply learnings from the case to future situations.

Each part of the rubric consisted of multiple subcategories, each assessed on a 10-point scale, ranging from “Inadequate” to “Excellent”. This nuanced grading system ensured that the student’s understanding and application of the case material was accurately reflected.

However, a key aspect of this grading system was its flexibility. If any criteria were not applicable in a given case, they were omitted from the grading, ensuring that the evaluation was always relevant to the exercise at hand. This consideration was a reflection of the diverse and unpredictable nature of real-world clinical scenarios, underscoring the flexibility required in both the AI’s grading approach and in the students’ learning process.

The development and implementation of this AI chatbot system illustrates the immense potential of artificial intelligence to transform and enhance pharmacy education. By creating a responsive facilitator that guides students through the nuances of a patient case, we aimed to foster an interactive, engaging learning environment conducive to analytical thinking and clinical application. The AI was engineered not just to present information, but to cultivate the critical thought processes that empower students to make sound clinical judgments as future pharmacists. Moving forward, intelligent systems like this chatbot can continue to push the boundaries of technology-enabled education, ensuring that the next generation of healthcare professionals are equipped with the skills needed to provide optimal patient care. As AI capabilities expand, so too will the ability to deliver personalized, adaptive learning experiences that make students active drivers in their educational journey.