Clinical Informatics Certification Clinical Informatics Data Analytics 3 — Questions and Answers
Question 1: A clinical analyst creates a dashboard showing monthly ED visit counts. Which chart type best displays trends over time?
- Pie chart
- Box plot
- Line chart (Correct answer)
- Heat map
Correct answer: Line chart
Line charts are ideal for displaying continuous trends over time, making them the standard choice for time-series data.
Question 2: Which term describes a variable that is correlated with both the exposure and the outcome, potentially distorting study results?
- Mediator
- Moderator
- Confounder (Correct answer)
- Covariate
Correct answer: Confounder
A confounder is independently associated with both the exposure and the outcome, creating a spurious or masked association.
Question 3: In survival analysis, what does the Kaplan-Meier curve represent?
- The probability of an event occurring over time in a cohort (Correct answer)
- The sensitivity and specificity of a diagnostic test
- The linear relationship between two continuous variables
- The distribution of lab values across a population
Correct answer: The probability of an event occurring over time in a cohort
The Kaplan-Meier curve estimates the survival function, showing the probability that subjects survive (or are event-free) beyond each time point.
Question 4: Which method allows a data scientist to assess a predictive model's generalizability by repeatedly partitioning data into training and testing sets?
- Bootstrapping
- K-fold cross-validation (Correct answer)
- Propensity score matching
- Sensitivity analysis
Correct answer: K-fold cross-validation
K-fold cross-validation splits data into k subsets and iterates so each subset serves as a test set once, reducing overfitting bias.
Question 5: Which FHIR resource is most appropriate for representing a patient's laboratory test result?
- Condition
- Observation (Correct answer)
- Procedure
- DiagnosticReport
Correct answer: Observation
The FHIR Observation resource captures individual measurements and lab results such as serum glucose or hemoglobin A1c.
Question 6: What is 'feature engineering' in the context of clinical predictive modeling?
- Selecting hardware infrastructure for model deployment
- Transforming and creating input variables to improve model performance (Correct answer)
- Auditing model outputs for regulatory compliance
- Publishing model results in peer-reviewed journals
Correct answer: Transforming and creating input variables to improve model performance
Feature engineering involves creating, selecting, or transforming predictor variables to help machine learning algorithms learn more effectively.
Question 7: Which bias occurs when clinicians order additional tests because they know a patient's outcome, influencing how the dataset is structured?
- Selection bias
- Measurement bias
- Surveillance bias (Correct answer)
- Recall bias
Correct answer: Surveillance bias
Surveillance bias (or detection bias) occurs when the intensity of monitoring differs between groups, leading to differential ascertainment of outcomes.
