A healthcare organization plans to develop an AI-driven diagnostic tool.

CPMAI’s Data Understanding and Data Preparation phases stress that AI success in domains like healthcare depends on robust data pipelines that ensure consistency, quality, and accessibility before modeling begins. Guidance describes these phases as profiling and assessing data, then performing cleaning, transformation, and structuring so that data are reliable and usable by downstream models.

A data quality assessment combined with ETL (extraction, transformation, loading) processes directly supports these objectives. ETL pipelines standardize formats across disparate systems, enforce validation rules, manage missing values, harmonize coding schemes (for example, diagnosis codes), and centralize data into accessible stores. This is exactly the kind of foundational work CPMAI describes as a prerequisite to effective model development, particularly in regulated sectors such as healthcare where inconsistent or inaccessible data can have clinical and regulatory consequences.

By contrast, using NLP to standardize records (B) is a specialized technique that may help later but does not replace a systematic quality and ETL process. Integrating EHR with ML algorithms (C) and designing hybrid cloud storage (D) are more about later technical integration and infrastructure than about defining and ensuring initial data consistency and accessibility. Thus, in line with CPMAI’s data-centric guidance, performing a data quality assessment with ETL processes is the correct method, making option A the best answer.