How Integrated AI Activities Impact Clinical Workflow Efficiency in Health Care

Clinical application of artificial intelligence (AI) has evolved from stand-alone applications focused on a specific activity. Today's advancements involve the integration of several activities in various stages of the clinical care process. The significance of this evolution lies in the fact that most inefficiencies found in clinical care occur due to the transitions between different clinical activities, e.g., intake, charting/documentation, coding, scheduling, and post-visit follow-up.

Why Fragmented Administrative Activity Remains an Organizational Issue

Fragmentation continues to influence clinical productivity. Time allocation, personnel deployment, and delivery of care are influenced by administrative burden. Documentation, claims preparation, and other administrative tasks occupy large segments of working time in many clinical settings.

However, the primary cause of fragmentation is the fact that each administrative function is typically performed using a different system/workflow. Therefore, the user must enter and verify data in each of the required systems before they can transfer the information to another system. The increased probability of errors, delay in service, and omission of important details further complicate this problem.

Insurance verification, prior approval, and appointment coordination also require the use of structured data; however, when these procedures are separated and disconnected, users must perform these bridging actions themselves. This limits the efficiency of the process and contributes additional variables to routine services.

What Do Multi-Role AI Systems Change in Practice?

Systems employing multiple roles for AI enable organizations to structure their tasks as part of a continuous workflow instead of individual functions. For instance, rather than having separate tools for documentations, coding, and front-desk services, multi-role systems link each role together within a shared framework.

Sully.ai is a model for integrating clinical and administrative support with multiple roles. The roles include but are not limited to scribes who create documentation based upon what was said during patient consultations; coders who translate clinical documentation into standardized language for purposes of billing and reimbursement; and patient communications. All three of these roles are connected via common data flows. Data collected during one phase of the service is available to subsequent phases without being re-entered by the user.

The benefit is operational as opposed to technical. While these systems do replace some aspects of human expertise, they do so to eliminate redundant human activity and minimize opportunities for degradation or misunderstanding of patient information.

System-Level Effects and Workflow Implications

From a workflow standpoint, the implementation of multi-role AI impacts how tasks are assigned and completed. Most routine administrative work can be done continuously in the background while clinical staff are able to concentrate on making decisions and interacting with patients.

There are tangible effects from an operational standpoint. Reducing the number of manual hand-offs decreases processing time. Providing standardized data flow enables more consistency across similar types of cases. Decreasing the number of manual hand-offs can lead to reduced error rates associated with transcription/duplication.

On the other hand, implementing multi-role AI introduces new dependency relationships. Any failure in one component may have negative impact on downstream components. Designing and maintaining systems is therefore more critical than ever before.

As always, interoperability is a major factor. If multi-role AI systems cannot communicate with existing electronic health records or billing systems, then the potential advantages will be diminished. Whether or not interoperability minimizes complexity or merely moves it to another area will depend on how well the systems are designed to interact with each other.