Deepthi Bathina, Founder and CEO of GW RhythmX.
getty
AI won’t transform healthcare simply by generating more insights. It will create value when those insights are connected to the workflows, teams and actions that ultimately improve patient outcomes.
Healthcare leaders are already seeing AI entering scheduling, documentation, risk prediction, quality improvement, capacity management and revenue cycle workflows. The harder question is more practical: How should leaders decide which AI investments will actually improve care delivery, rather than add another layer of operational complexity?
This is where care orchestration becomes critical. Consider a recently discharged heart failure patient. AI may identify a high risk of readmission based on missed medications, transportation barriers and a lack of follow-up care. The insight is valuable, but it does not reduce readmission risk on its own. Outcomes improve only when the right team member receives that information, intervenes and helps the patient complete the next step in care. The difference between intelligence and orchestration is execution.
Start with the workflow, not the tool.
Specialized AI applications can be valuable. One may help identify patients at risk of missing appointments, while another surfaces gaps in care, and others support documentation, staffing or prior authorization.
The problem emerges when each tool creates its own destination for attention. A new dashboard, queue or alert may look useful in isolation, but it can become one more place for busy teams to check. Over time, that can make a health system more informed without making it more coordinated.
This problem is not solved by simply embedding an app in the electronic health record (EHR). Solutions must truly break down silos in data, decisions and processes if they are going to deliver value. Otherwise even a well-integrated tool is just adding clicks and cognitive burden to an already-strained system.
Before adopting or expanding an AI capability, leaders should map the workflow it is meant to influence. Returning to the heart failure example, an organization should be able to answer a series of practical questions before judging the AI successful. Does the alert go to a care manager, nurse navigator or primary care team? Is the first intervention a phone call, text message, transportation referral, medication assistance program or some combination of those actions? If the patient cannot be reached, when is the case escalated and to whom? The answers matter more than the predictive score itself because they determine whether the insight produces action.
Build care intelligence around decisions.
Care intelligence creates value when it helps the right person make the right decision at the right time. To make that happen, leaders should define the decision before they define the metric. A readmission-risk model, for example, is most useful when the organization has already determined what action should follow, who owns that action and how it will be tracked.
A practical test is to ask whether the insight changes what someone does next. If the insight doesn’t change a decision, trigger a workflow, engage a patient or escalate a risk, it may be informative, but it’s not yet operational. If the answer is unclear, the organization may have a reporting capability, but not yet an operating capability.
This distinction is especially important as health systems pursue value-based care. Success depends on identifying risk, as well as coordinating timely, appropriate interventions across clinical, operational and administrative teams.
That coordination is difficult to achieve if intelligence remains trapped in separate applications or limited to individual users instead of being shared across the full care team.
Make data and interoperability part of the strategy.
Delivering precision care requires more than analytics. It requires understanding the complete context around a patient and the actions needed to move care forward.
Too often, healthcare AI discussions focus on the EHR as the primary source of truth. In reality, critical information about a patient and the care they need exists across many systems and teams. Clinical history may reside in the EHR, while health plan policies, social determinants data and referral networks live elsewhere. Organizations rely on institutional knowledge when determining what action should happen next.
For AI to create meaningful value, it must understand both the clinical and operational context surrounding a patient.
Interoperability is not simply about moving data between applications. It’s about creating a complete patient context that enables informed decisions and effective action.
Healthcare has spent decades building systems of record. The next opportunity is building systems of action that connect intelligence, workflows and accountability around the patient journey.
Rather than adding more technology to an already complex system, the real opportunity is to help patients, clinicians and care teams move through care with greater clarity, coordination and confidence. For health systems evaluating their next AI investments, the most important question is simple: Does this help the organization understand the full patient context and follow through on the actions that improve outcomes?
Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?
