How AI Chart Review Saves Your Coders—And Your Bottom Line

More than half of your Medicare patients are in a risk-adjusted payment model right now. According to KFF, Medicare Advantage enrollment reached 35.2 million beneficiaries in 2026 — covering 55% of all eligible Medicare recipients. That means HCC coding accuracy is not a billing department problem. It is your primary revenue lever.

The rules just changed permanently. The CMS-HCC V28 model is fully live at 100% for payment year 2026, and according to ChiroKHealth, it eliminated over 2,000 ICD-10-CM codes from the HCC set — putting practices that haven't updated their documentation at risk of RAF score drops between 10% and 20%.

At the same time, CMS expanded RADV audits from roughly 60 MA contracts per year to all 550 eligible contracts, running quarterly. The DOJ drove the point home with a $556 million settlement over invalid MA diagnosis codes in January 2026, per RAAPID Inc. This is no longer an abstract compliance risk.

AI-powered chart review is the operational response internal medicine practices need — and it solves two problems simultaneously: a coder workforce in crisis and a compliance environment that demands defensible, evidence-linked documentation on every single code.

What You Need to Know About AI Chart Review, Coder Burnout, and Internal Medicine Practices

Internal medicine practices are being compressed from two directions at once — and both are getting worse.

On the compliance side, the FCA Working Group has explicitly identified Medicare Advantage fraud as a priority enforcement area. According to Doctors Management's OIG guidance analysis, federal False Claims Act recoveries hit $6.8 billion in fiscal year 2025, driven substantially by MA-related cases. Unsupported HCC codes are no longer an audit inconvenience — they are a federal liability.

On the workforce side, the people responsible for catching those codes are burning out and leaving.

• Physician burnout held at 41.9% in 2025, per the AMA's Organizational Biopsy of nearly 19,000 physicians — but documentation burden and EHR fatigue remain the unresolved structural drivers
• The Medscape 2024 Physician Burnout Report, cited by the Society of General Internal Medicine, found IM physicians spending more than two hours daily outside work hours on administrative tasks — with HCC documentation cited as a major contributor
• According to ChiroKHealth citing AAPC data, the national shortage of certified medical coders sits at 30%, with more than 90% of independent practices reporting high coder turnover

Manual chart review is the thread connecting both crises. It exhausts coders, introduces documentation gaps, and leaves practices exposed when auditors arrive.

AI chart review tools like ForeSee ESP exist precisely to cut that thread — surfacing evidence-linked diagnoses automatically, so coders validate rather than hunt.

Key Strategies and Best Practices

The average internal medicine practice is missing 8–12 HCC coding opportunities per 100 Medicare Advantage patients, according to OmniMD citing the National Association of ACOs. That is not a documentation problem — it is a workflow problem. Manual chart review at scale is not sustainable, and the numbers prove it.

Three practice-level strategies close that gap immediately:

• Automate chart hunting, not clinical judgment. ForeSee Medical's ESP platform uses NLP to surface conditions buried in free-text notes and PDF records — then links each suspect diagnosis to exact source evidence via its InstaVu® function. Coders validate; they stop hunting.
• Treat RAF score accuracy as a revenue line item. According to Invene, every 0.1 increase in RAF score generates approximately $1,040 in additional annual revenue per MA member. For a practice with 200 MA patients, closing even half the coding gap is a six-figure revenue event.
• Staff retention starts with removing the worst part of the job. ChiroKHealth reports that 31% of coding staff are actively considering leaving — overwhelmingly driven by manual documentation overload. AI-assisted review removes that cognitive burden before it becomes a resignation.

As Sol Lizerbram, DO, Executive Chairman and Co-Founder of ForeSee Medical, put it: "Using artificial intelligence in healthcare is the future to financial success and quality of care in the risk-adjusted payment model." That is not a vision statement anymore — it is an operational requirement.

Common Mistakes to Avoid

Most practices deploying AI coding tools make the same error: they treat the software as a replacement for clinical judgment rather than an accelerant for it. The result is sloppy validation, unsupported codes, and exactly the kind of audit exposure they were trying to escape.

RAAPID Inc. is direct on this point: "Defensible, encounter-linked coding is the new bar — accuracy alone is no longer sufficient." That distinction matters. A flagged diagnosis without MEAT criteria in the chart is a liability, not revenue.

The other common failure is choosing tools that haven't been independently vetted. Not all AI coding platforms carry third-party security certification — and under expanded RADV audit scrutiny, that gap is not theoretical.

• Skipping clinical validation: AI surfaces suspects. A credentialed coder or physician still confirms them. Bypassing this step invites denials and audit findings.
• Ignoring the Delete Suspects step: Submitting last year's diagnoses without current-year clinical evidence is a primary RADV trigger.
• Selecting uncertified platforms: According to Yahoo Finance / BusinessWire, ForeSee ESP achieved HITRUST r2 Certification in July 2026 — the benchmark for practices that cannot afford a data breach on top of an audit.
• Treating AI as a one-time fix: Per Inflo Health citing Pelletier et al., JAMIA Open, burnout dropped from 54.9% to 33.3% post-AI implementation — but only when the tool was integrated into the daily workflow, not deployed and abandoned.

AI reduces burden when it removes work. Practices that bolt tools onto broken workflows don't capture that benefit — they add complexity to chaos.

Practical Next Steps

CMS expanded its RADV audit coder workforce from 40 to an intended 2,000 certified coders — and deployed AI tools to do it faster, according to Wolters Kluwer. If the federal government is using AI to audit your charts, your practice needs AI to prepare them.

Here is where to focus first:

• Audit your current HCC workflow. If your coders are spending 45+ minutes per chart hunting for diagnoses manually, that is your baseline problem. RAAPID Inc. benchmarks show AI-assisted review cuts that to under 10 minutes at 98%+ accuracy.
• Address unsupported codes immediately. ForeSee ESP's Delete Suspects report flags previously captured diagnoses that lack current clinical evidence — your first line of defense before an auditor finds them.
• Use InstaVu® to close the evidence gap. Every diagnosis recommendation links directly to the source chart entry, giving coders defensible, click-to-verify documentation.
• Stop letting denied claims die. According to Innovaccer, more than half of coding-error denials are never resubmitted — that is permanent revenue loss your practice cannot afford under V28.

According to Black Book Research, 92% of senior payer executives now consider advanced analytics and AI tools critical for accurate risk scoring and compliance. Your payer counterparts are already operating at this level.

Internal medicine practices facing Medicare Advantage contracts have two problems converging at once: a compliance environment that punishes coding errors with federal enforcement action, and a coder workforce that is stretched too thin to keep up with manual chart review demands. Neither problem is going away. Together, they are squeezing practice revenue from both directions simultaneously.

The answer is not hiring more coders. With a 30% national shortage of certified medical coders, according to ChiroKHealth citing AAPC, that market simply does not supply what you need. The answer is changing what your coders are doing — replacing manual chart hunting with AI-surfaced, evidence-linked code suggestions that are already mapped to the specific documentation required to survive a RADV audit.

ForeSee ESP's Delete Suspects report is a practical example of what this looks like operationally: the platform proactively flags previously captured diagnoses that now lack current clinical evidence, protecting practices from submitting codes that cannot withstand audit scrutiny. That is not a productivity feature. That is audit armor.

The practices that will absorb V28's RAF score compression and emerge with stronger revenue are the ones treating AI chart review as an operational standard right now — not a future consideration.

Contact Perfect Rhythm to see how ForeSee ESP fits your Medicare Advantage panel.

Frequently Asked Questions

How does AI chart review reduce coder burnout in internal medicine practices?

AI chart review tools like ForeSee ESP scan patient records using NLP and surface clinically documented conditions with exact source evidence already identified — eliminating the manual chart hunting that drives cognitive fatigue. Coders shift from searching to validating, which reduces workload and improves job satisfaction.

What is the connection between HCC coding accuracy and Medicare Advantage revenue?

In Medicare Advantage, reimbursement is tied directly to RAF scores, which are built from HCC codes. According to Invene, every 0.1 increase in RAF score translates to approximately $1,040 in additional annual revenue per MA member — making coding accuracy a direct financial lever, not an administrative detail.

Is ForeSee ESP compliant with the CMS-HCC V28 model?

Yes. ForeSee ESP is built for V28 compliance, addressing the elimination of roughly 2,000+ ICD-10-CM codes from the HCC set that created RAF score drops for practices still using V24-era documentation workflows. The platform is also HITRUST r2 certified, confirming independent third-party cybersecurity validation.

What does a RADV audit actually look for?

According to RAAPID Inc., RADV auditors require every diagnosis submitted for risk adjustment to be supported by a face-to-face encounter with MEAT criteria documented in the medical record. Vague language, missing provider signatures, or unsupported codes are sufficient grounds to disallow a code and claw back payment.

Can a small internal medicine practice realistically implement AI chart review?

Yes — and the smaller the practice, the higher the relative risk of NOT doing it. Practices with fewer coders absorb the most damage from turnover and manual workflow inefficiency. AI-assisted platforms that integrate with existing EHR systems require no new staff and are specifically designed to scale down to independent and small-group practice environments.