Gen AI is reshaping how Aetna handles its annual HEDIS review, a task that once demanded thousands of work weeks to scan more than 10 million medical records for gaps in care.
AI-driven platform cuts manual effort by two‑thirds
The health insurer has rolled out a document‑intelligence system that automatically extracts relevant clinical data from unstructured charts. According to Nathan Frank, Aetna’s chief digital and technology officer, the tool has slashed the need for manual review by about 65 percent.
Industry benchmarks suggest a full manual process would require roughly 50,000 work weeks—equivalent to nearly 1,000 full‑time staff. He noted that a team of 50 reviewers would need more than two decades to finish a single annual review without automation.
In a six‑month sprint, his team built a proof‑of‑concept that could process millions of records in two weeks. The AI-driven platform, named AI Medical Chart Review, leverages cloud services and large language models to decipher handwritten notes, pull diagnosis codes, medication lists, lab results and visit details, then rank records by the likelihood of closing a HEDIS measure.
“Large language models and gen AI give us the ability to train a model to decipher the charts, identify the high value codes, and build correlations,” he said. The system has already handled 14 million documents, and Aetna earned a CIO 100 Award for the innovation.
Related: Blacklisted Chinese firms still buy AI through Singapore
Human reviewers shift to quality control
After the AI screens and prioritizes records, human coders validate the findings and ensure the automated process works as intended. He explained that the reduced manual workload is being redirected toward quality‑control tasks and other higher‑value activities.
The platform’s speed has boosted gap‑closure rates, which in turn improves Star Ratings and increases reimbursement from payers. Streamlined workflows also free up staff who previously spent days or weeks on repetitive chart checks.
One practical challenge the AI tackles is the “white space filled with handwritten notes” that typify many medical charts. By converting that noisy input into structured data, the system helps Aetna meet the performance measures set by the National Committee for Quality Assurance, the body that maintains the HEDIS framework since 1991.
Frank highlighted the collaborative design approach that underpinned the project. Rather than drafting exhaustive specifications, engineers worked side‑by‑side with subject‑matter experts throughout the build. “It allowed us to move much faster, and having a business subject matter expert sitting in the same virtual or physical room with us got us a much better outcome,” he said.
