Atlantic Health System's AI approach to chronic disease control: measurable gains, lighter workloads
Atlantic Health System serves New Jersey, Pennsylvania, and the New York metro area with more than 550 sites of care, eight hospitals, and a clinically integrated network of 2,500+ providers. Their ACO manages 500,000+ lives across 11 value-based contracts. Despite strong infrastructure and outcomes, leaders still saw too many uncontrolled diabetics and hypertensives, and care teams stretched thin.
The mandate was clear: improve control, protect clinician time, and create a better patient experience-without adding headcount. The team moved beyond practice redesigns and analytics to deploy an AI-based clinical management platform called Medical Brain.
The management problem
Providers had limited time for visits and almost no bandwidth for non-visit outreach. Population health teams were solid, but scaling personal engagement was hard. Internal analytics flagged large pockets of persistently uncontrolled diabetes and hypertension-risk that would jeopardize quality scores and downstream costs.
Leaders committed to the quadruple aim, including clinician well-being. That meant solutions had to reduce friction for providers, not add more clicks or inbox noise.
The approach: AI that works for providers and patients
Medical Brain aggregates data from the EHR and other sources, then generates clinical decision support using thousands of evidence-based guidelines. Insights go to providers and care teams-and directly to patients through a mobile app with an AI chat assistant.
Patients get precision recommendations, education, and 24/7 support for symptoms, medications, and chronic conditions. Practices get fewer inbound messages and clearer next steps. The ACO targeted three goals: better clinical control (diabetes and hypertension), improved prevention/wellness, and a stronger patient experience-while offloading routine communication from providers.
Rollout: start voluntary, prove value, expand
The ACO invited interested practices to opt in. Data extracts were automated from EHRs; teams co-designed use cases and workflows. In parallel, transition-of-care nurses used the platform to reduce readmissions.
Independent practices adopted quickly due to time savings and patient feedback. Employed and joint-venture groups focused on which use cases to shift from the EHR portal and how to embed the platform into daily workflows. As with any new tool, there were early adopters, cautious adopters, and wait-and-see groups-the difference-maker was visible impact.
Outcomes that matter to leaders
Quality measures were aligned to CMS definitions for diabetes control and hypertension control. Once enough patients engaged with the platform, Atlantic Health ran pre/post analyses.
- Diabetes (A1c): Among patients starting with A1c >9%, 64% achieved control after using the platform. Those who moved into control saw an average A1c drop of 3.66 points. Across all with A1c >9%, the average reduction was 2.52 points; for those >8%, it was 1.8 points. See CMS guidance: HbA1c Poor Control (>9%).
- Hypertension (BP): Patients with average systolic ≥160 dropped by 34 points; >150 dropped by 20; >140 dropped by 12; >130 dropped by 5. CMS BP control measure: Controlling High Blood Pressure.
- Inbox relief: Only 6.4% of patient chats required escalation to providers; 93%+ were resolved by the AI assistant and care team workflows. In a nephrology practice with medically complex patients, escalations were just 1.3%.
The takeaway: thousands of patients moved closer to control, with fewer ER visits and hospitalizations expected, while clinicians reclaimed time and focus.
Why adoption stuck
- No extra staffing: The platform augmented existing teams instead of adding cost.
- Provider-first experience: Recommendations surfaced inside workflows; low manual effort.
- Patient utility: 24/7 answers, clear next steps, and education built trust and usage.
- Data leverage: Real-time EHR extracts plus other sources made outreach relevant.
- Clear wins: Measurable reductions in A1c and BP-and message deflection-built momentum.
Implementation playbook for health system leaders
- Start with a focused population: High-risk diabetics and hypertensives give fast signal on clinical outcomes and cost avoidance.
- Co-design workflows: Let practices pick initial use cases (e.g., med questions, symptom triage, refills) and clarify routing rules for escalation.
- Integrate where people work: Surface insights and patient status inside the EHR and care team tools.
- Run impact analyses early and often: Pre/post on A1c, BP, AWVs, screenings, and message volumes. Share results transparently to drive adoption.
- Protect clinician time: Use AI chat to buffer low-value messages and standardize responses to common questions.
- Expand in stages: After diabetes and hypertension, move to broader quality metrics and medication adherence.
What's next at Atlantic Health
The ACO is scaling Medical Brain as standard of care for diabetes and hypertension, then pushing into the full population health playbook: adherence, screenings, wellness, behavioral health, social factors, even spiritual needs. The goal is simple-better control, better experience, fewer avoidable costs, and less burnout.
For managers building AI capability
If your organization is upskilling leaders and teams to apply AI in care operations, explore role-specific learning paths: AI courses by job.
Your membership also unlocks:
