Verbal Reasoning Quiz
20 questions · no time limit
0:00
READING COMPREHENSION
Question 1 of 20
No time limit
Current
Answered
Flagged
Not seen
Read the passage and answer the question.
Because machine-learning models can detect patterns in massive datasets, some hospitals have begun using them to predict which patients are at high risk of readmission. Early reports touted impressive accuracy, but subsequent audits found that some models achieved those results by relying on proxies for socioeconomic status, such as zip code or insurance type. When those proxies are embedded in predictions, clinicians may inadvertently allocate extra attention to patients already advantaged by easier access to follow-up care.
One proposed remedy is to constrain models so that predictions are statistically independent of certain sensitive attributes. Yet independence constraints can be misleading in medicine: if a disease genuinely has different prevalence across populations due to environmental exposure, forcing identical prediction rates can hide real risk. A more promising approach, some researchers argue, is to pair predictive tools with "impact evaluations" that measure how the tool changes clinical decisions and outcomes across groups.
The primary function of the second paragraph is to:
YOUR PROGRESS
20
QUESTIONS
Ready when you are
Answer in any order. Your progress shows here as you go.
QUESTIONS
KEYBOARD
Pick an answer ABCDE
Next question ↵→
Previous ←
Flag for review F
Answers save as you go · flagged questions come back at the end