MCQ Quiz-Artificial Intelligence and Machine Learning in Pharmacy

MCQ Quiz: Artificial Intelligence and Machine Learning in Pharmacy

by

Artificial Intelligence (AI) and its subfield, Machine Learning (ML), are transforming healthcare by enabling computers to learn from data, identify patterns, and make predictions. For pharmacists, these technologies are not just futuristic concepts but are actively being integrated into clinical decision support, drug discovery, and population health management. This quiz for PharmD students will test your understanding of the foundational principles of AI/ML and their growing role in advancing the practice of pharmacy.

1. “Machine Learning” is best defined as a field of AI that:

  • Focuses on building robots that look and act like humans.
  • Gives computer systems the ability to learn from data without being explicitly programmed for every scenario.
  • Involves the manual writing of complex “if-then” rules for every possible clinical situation.
  • Is concerned only with the physical storage of large datasets.

Answer: Gives computer systems the ability to learn from data without being explicitly programmed for every scenario.

2. A key requirement for training a successful and accurate clinical machine learning model is:

  • A small, simple dataset.
  • A large, high-quality, and well-labeled dataset.
  • The personal opinions of a few expert clinicians.
  • A very fast internet connection.

Answer: A large, high-quality, and well-labeled dataset.

3. In a “supervised learning” model, the algorithm is trained on data that has been:

  • Left completely raw and unorganized.
  • Anonymized to protect patient privacy.
  • Labeled with the correct outcomes or answers.
  • Sourced from social media.

Answer: Labeled with the correct outcomes or answers.

4. An AI model that analyzes thousands of patient charts to identify previously unknown risk factors for an adverse drug reaction is an example of:

  • A simple rule-based alert.
  • “Unsupervised learning” to find hidden patterns in data.
  • A manual data review.
  • A basic dispensing function.

Answer: “Unsupervised learning” to find hidden patterns in data.

5. “Deep learning” is a more advanced type of machine learning that uses:

  • Simple statistical regressions.
  • Decision trees.
  • Multi-layered neural networks, inspired by the structure of the human brain.
  • A set of pre-programmed clinical rules.

Answer: Multi-layered neural networks, inspired by the structure of the human brain.

6. A pharmacist uses an AI-powered tool to analyze a patient’s EHR data and predict their 5-year risk of a cardiovascular event. This is an application of:

  • Descriptive analytics.
  • Predictive analytics.
  • A standard drug interaction check.
  • Inventory management.

Answer: Predictive analytics.

7. “Natural Language Processing” (NLP) is a branch of AI that enables computers to:

  • Understand, interpret, and generate human language, such as extracting information from a clinician’s free-text note.
  • Perform complex mathematical calculations.
  • Process images like X-rays or CT scans.
  • Dispense medications accurately.

Answer: Understand, interpret, and generate human language, such as extracting information from a clinician’s free-text note.

8. The problem of “algorithmic bias” is a major ethical concern in AI because a model trained on biased data can:

  • Learn to be perfectly fair and equitable.
  • Make systematically prejudiced predictions that worsen health disparities for minority populations.
  • Correct its own biases without human intervention.
  • Refuse to make any predictions for certain patient groups.

Answer: Make systematically prejudiced predictions that worsen health disparities for minority populations.

9. The use of data standards like RxNorm (for drugs) and ICD-10 (for diagnoses) is critical for machine learning because they:

  • Provide the structured, consistent data needed to train and run the models effectively.
  • Make the data more difficult for a computer to understand.
  • Are not relevant to AI and ML.
  • Are only used for billing, not for clinical analysis.

Answer: Provide the structured, consistent data needed to train and run the models effectively.

10. A key role for the pharmacist in the era of AI and ML will be:

  • To become a data scientist and write all the algorithms.
  • To act as a “clinical interpreter,” validating the output of AI models and applying the insights to individual patient care.
  • To focus solely on manual, repetitive dispensing tasks.
  • To resist the implementation of all new technologies.

Answer: To act as a “clinical interpreter,” validating the output of AI models and applying the insights to individual patient care.

11. An AI model that analyzes retinal scans to detect diabetic retinopathy is an application of which branch of AI?

  • Natural Language Processing (NLP)
  • Robotics
  • Computer Vision
  • Predictive Analytics

Answer: Computer Vision

12. The “black box” problem of some complex AI models refers to the difficulty in:

  • Understanding the exact reasoning or features the model used to arrive at a specific prediction.
  • Storing the model securely.
  • Getting the model to work on a standard computer.
  • Training the model with new data.

Answer: Understanding the exact reasoning or features the model used to arrive at a specific prediction.

13. In the field of drug discovery, machine learning is used to:

  • Speed up the process by predicting the properties of new molecules and identifying potential drug candidates.
  • Manually test thousands of chemical compounds in a lab.
  • Write the final marketing materials for a new drug.
  • Set the price for a new medication.

Answer: Speed up the process by predicting the properties of new molecules and identifying potential drug candidates.

14. A machine learning model designed to predict medication non-adherence would be trained on which type of data?

  • Prescription fill history.
  • Patient demographics.
  • Social determinants of health data.
  • All of the above.

Answer: All of the above.

15. A key difference between AI and traditional pharmacy automation is that AI:

  • Is designed to perform fixed, repetitive tasks, while automation is designed to learn and adapt.
  • Is designed to learn and adapt, while traditional automation performs fixed, repetitive tasks.
  • Has no role in pharmacy operations.
  • Is less expensive than traditional automation.

Answer: Is designed to learn and adapt, while traditional automation performs fixed, repetitive tasks.

16. The concept of “garbage in, garbage out” is especially true for machine learning, meaning:

  • The quality of the output or prediction is highly dependent on the quality of the input data.
  • All data, regardless of quality, will produce a good result.
  • The model can fix errors in the input data automatically.
  • The amount of data is more important than its quality.

Answer: The quality of the output or prediction is highly dependent on the quality of the input data.

17. The use of a “training set,” “validation set,” and “test set” is a standard practice in machine learning to:

  • Make the model more complex.
  • Develop a model and then test its performance on unseen data to ensure it generalizes well.
  • Use all available data for training the model.
  • Comply with a specific legal requirement.

Answer: Develop a model and then test its performance on unseen data to ensure it generalizes well.

18. A pharmacist’s knowledge of _________ is essential for identifying potential biases in the data used to train a clinical AI model.

  • Health disparities and social determinants of health
  • Sterile compounding procedures
  • The chemical structure of drugs
  • Pharmacy law

Answer: Health disparities and social determinants of health

19. A “chatbot” powered by AI could be used in a pharmacy setting to:

  • Answer common patient questions about medication administration or side effects.
  • Replace the need for a pharmacist to verify prescriptions.
  • Provide counseling on complex clinical cases.
  • Manage the pharmacy’s inventory.

Answer: Answer common patient questions about medication administration or side effects.

20. The “human-in-the-loop” principle is critical for the safe use of clinical AI and ML, asserting that:

  • The AI should have the final say in all clinical decisions.
  • The human clinician is the ultimate decision-maker who uses the AI as a tool.
  • Humans are only involved in the initial programming.
  • The AI is designed to work without any human interaction.

Answer: The human clinician is the ultimate decision-maker who uses the AI as a tool.

21. A machine learning model could improve medication reconciliation by:

  • Automatically filling all of a patient’s prescriptions.
  • Analyzing and flagging potential discrepancies from multiple sources (EHR, claims data, patient report) for the pharmacist to review.
  • Making the process entirely manual.
  • Only using data from a single source.

Answer: Analyzing and flagging potential discrepancies from multiple sources (EHR, claims data, patient report) for the pharmacist to review.

22. How can machine learning enhance the practice of pharmacogenomics?

  • By identifying novel and complex gene-drug associations from large genomic and clinical datasets.
  • By making genetic testing obsolete.
  • By simplifying all genetic information into a single risk score.
  • It has no application in pharmacogenomics.

Answer: By identifying novel and complex gene-drug associations from large genomic and clinical datasets.

23. The role of a pharmacist on a team developing a new AI tool is to provide:

  • The data science expertise.
  • The subject matter expertise on the medication use process.
  • The project funding.
  • The marketing plan.

Answer: The subject matter expertise on the medication use process.

24. The FDA’s evolving regulatory framework for AI/ML in medical devices must account for the fact that:

  • The algorithms are “locked” and never change after approval.
  • The algorithms can learn and change over time, requiring a new approach to post-market surveillance.
  • These devices do not require any regulation.
  • The FDA does not regulate medical devices.

Answer: The algorithms can learn and change over time, requiring a new approach to post-market surveillance.

25. A pharmacist who is a “lifelong learner” is better prepared for a future with AI because they:

  • Will be able to adapt to new technologies and integrate them into their practice.
  • Can rely solely on the knowledge they gained in pharmacy school.
  • Will resist all technological change.
  • Believe that AI will not impact pharmacy.

Answer: Will be able to adapt to new technologies and integrate them into their practice.

26. A key difference between AI and human intelligence is that AI currently excels at:

  • Empathy and common-sense reasoning.
  • Pattern recognition in massive datasets.
  • Creative and abstract thought.
  • Understanding nuanced social contexts.

Answer: Pattern recognition in massive datasets.

27. An AI model that analyzes a patient’s “digital phenotype” (data from their smartphone and wearables) might be able to:

  • Diagnose a medical condition with 100% certainty.
  • Predict early signs of health deterioration, such as a worsening of heart failure.
  • Prescribe medications automatically.
  • Replace the need for an EHR.

Answer: Predict early signs of health deterioration, such as a worsening of heart failure.

28. A pharmacy leader who is “forging ahead” would view the implementation of AI and ML as a(n):

  • Threat to the profession.
  • Opportunity to enhance patient care and pharmacist capabilities.
  • Unnecessary expense.
  • Distraction from core dispensing functions.

Answer: Opportunity to enhance patient care and pharmacist capabilities.

29. The use of _________ is essential for training a machine learning model to be useful in a clinical workflow.

  • Clean, structured, and interoperable data
  • Messy, unstructured data only
  • A small, biased dataset
  • Data from a single, healthy patient

Answer: Clean, structured, and interoperable data

30. The ultimate validation of a clinical machine learning model is:

  • Its performance on a retrospective dataset.
  • Its publication in a prestigious journal.
  • Its demonstrated ability to improve patient outcomes in a prospective, real-world clinical setting.
  • The complexity of its algorithm.

Answer: Its demonstrated ability to improve patient outcomes in a prospective, real-world clinical setting.

31. How can AI and ML help to combat “alert fatigue”?

  • By creating more, less specific alerts.
  • By learning which alerts are clinically meaningful for a specific patient and context, and suppressing the “noise.”
  • By turning off the CDS system entirely.
  • They cannot help with alert fatigue.

Answer: By learning which alerts are clinically meaningful for a specific patient and context, and suppressing the “noise.”

32. The term “algorithm” in machine learning refers to:

  • The data used to train the model.
  • The set of rules or statistical processes that the model uses to learn from data.
  • The user interface of the software.
  • The computer hardware.

Answer: The set of rules or statistical processes that the model uses to learn from data.

33. A key skill for pharmacists of the future will be “data literacy,” which is the ability to:

  • Read, understand, create, and communicate with data.
  • Write complex computer programs.
  • Repair hardware.
  • Manage a pharmacy’s social media.

Answer: Read, understand, create, and communicate with data.

34. The integration of AI into pharmacy automation could lead to:

  • Robots that can only perform a single, pre-programmed task.
  • “Smarter” robots that can adapt to changing workflows or identify incorrect products using computer vision.
  • A complete stop to all pharmacy automation.
  • An increase in the size of dispensing robots.

Answer: “Smarter” robots that can adapt to changing workflows or identify incorrect products using computer vision.

35. A significant challenge in using AI for clinical decision support is ensuring its recommendations are:

  • Difficult to understand.
  • Explainable and transparent.
  • Based on outdated information.
  • Kept secret from the clinician.

Answer: Explainable and transparent.

36. A pharmacist’s expertise in _________ is critical for determining if an AI model’s prediction is clinically plausible for a specific patient.

  • Biostatistics
  • Pathophysiology and pharmacology
  • Computer engineering
  • Medical billing

Answer: Pathophysiology and pharmacology

37. The “usability” of an AI-powered tool is crucial because:

  • A complex and confusing interface will prevent clinicians from using the tool safely and effectively.
  • It is not an important factor.
  • It determines the underlying accuracy of the algorithm.
  • A good user interface can make up for a biased algorithm.

Answer: A complex and confusing interface will prevent clinicians from using the tool safely and effectively.

38. The use of “big data” from EHRs to train machine learning models is an example of:

  • Primary data collection through a clinical trial.
  • Secondary use of clinical data.
  • A violation of HIPAA if not properly de-identified.
  • Both B and C.

Answer: Both B and C.

39. A pharmacy leader implementing a new AI system must consider the impact on:

  • Clinical workflows.
  • Staff training and education.
  • Patient safety.
  • All of the above.

Answer: All of the above.

40. The future role of the pharmacist is not to compete with AI, but to:

  • Resist its implementation at all costs.
  • Augment their own skills and intelligence with AI as a powerful tool.
  • Ignore it completely.
  • Let AI handle all patient interaction.

Answer: Augment their own skills and intelligence with AI as a powerful tool.

41. An example of “reinforcement learning,” a type of ML, would be:

  • Training a model to identify images of skin cancer from a labeled dataset.
  • An algorithm that learns to optimize a chemotherapy dosing regimen by getting “rewards” for positive outcomes and “penalties” for negative outcomes over many simulated trials.
  • Clustering patients into groups based on their clinical characteristics.
  • A simple linear regression model.

Answer: An algorithm that learns to optimize a chemotherapy dosing regimen by getting “rewards” for positive outcomes and “penalties” for negative outcomes over many simulated trials.

42. The “Introduction to Pharmacy Informatics” course provides the foundational knowledge that is essential for understanding:

  • How data is structured and used, which is a prerequisite for any AI or ML application.
  • The chemical basis of drug action.
  • The legal regulations for sterile compounding.
  • The pathophysiology of heart disease.

Answer: How data is structured and used, which is a prerequisite for any AI or ML application.

43. A significant legal and ethical question in AI is accountability. If an AI model contributes to a medical error, who is responsible?

  • The clinician who used the tool.
  • The developer of the AI model.
  • The hospital that implemented the system.
  • This is a complex and unresolved question.

Answer: This is a complex and unresolved question.

44. How does AI relate to the field of “human factors engineering”?

  • The design of the user interface for an AI tool is a critical human factors consideration to ensure it is used safely.
  • The two fields are unrelated.
  • Human factors is only about the physical design of hardware.
  • AI eliminates all human factors considerations.

Answer: The design of the user interface for an AI tool is a critical human factors consideration to ensure it is used safely.

45. An AI model could be used to optimize pharmacy inventory by:

  • Ordering the same amount of each drug every week.
  • Predicting future prescription demand based on historical patterns, seasonal trends, and local health data.
  • Asking the pharmacist what to order.
  • Ordering only the most expensive medications.

Answer: Predicting future prescription demand based on historical patterns, seasonal trends, and local health data.

46. A pharmacist’s critical thinking skills are essential when interacting with an AI recommendation because they must:

  • Consider the patient’s unique context, values, and goals, which the AI may not fully grasp.
  • Always accept the AI’s recommendation without question.
  • Manually verify the AI’s calculation.
  • Be able to explain the algorithm’s code.

Answer: Consider the patient’s unique context, values, and goals, which the AI may not fully grasp.

47. Forging ahead into a future with AI requires the pharmacy profession to be:

  • Resistant to change.
  • Proactive in education, advocacy, and defining the pharmacist’s role in an AI-assisted healthcare system.
  • Passive and wait for other professions to lead.
  • Focused on reducing the use of all technology.

Answer: Proactive in education, advocacy, and defining the pharmacist’s role in an AI-assisted healthcare system.

48. A machine learning model that analyzes a patient’s genetic and clinical data to predict their response to an antidepressant is a tool for:

  • Precision psychiatry.
  • General population health.
  • Infectious disease management.
  • Cardiology.

Answer: Precision psychiatry.

49. The most important role of a pharmacist on a team implementing a new AI-driven CDS is to ensure:

  • The tool is clinically safe and evidence-based.
  • The tool is as complex as possible.
  • The tool is liked by the IT department.
  • The tool increases the number of alerts.

Answer: The tool is clinically safe and evidence-based.

50. The integration of AI and machine learning into pharmacy practice is ultimately aimed at:

  • Making the pharmacist’s job more difficult.
  • Improving the efficiency, safety, and personalization of patient care.
  • Increasing the cost of all aspects of healthcare.
  • Replacing the entire pharmacy workforce.

Answer: Improving the efficiency, safety, and personalization of patient care.

Leave a Comment