IKS Health
2026
NBA - Next Best Action
An AI-powered decision support system for healthcare revenue cycle teams that synthesizes complex claim data into high confidence recommended actions in seconds, cutting encounter review time from 12 minutes to under 30 seconds.
Overview
What is Next Best Action?
NBA is an AI-powered system embedded within healthcare Revenue Cycle Management (RCM) operations. When an analyst is reviewing a denied insurance encounter, the system instantly synthesizes all claim data, clinical codes, filing deadlines, and payer policies to surface a single, high-confidence recommended action.
Instead of forcing analysts to manually inspect root causes, clinical notes, and documentation requirements across multiple disjointed screens, NBA eliminates cognitive fatigue by telling them exactly what to do next - and why.
Through prioritized sticky action banners, transparent confidence scoring, and progressive disclosure, the design cuts encounter review time from 12 minutes to under 30 seconds while achieving a 96% resolution rate.
Problem
Too much data, not enough clarity
Healthcare operations teams process thousands of denied encounters every day. Each claim contains sprawling fields - denial codes, clinical analysis, payer data, filing deadlines - all scattered across multiple screens and disconnected systems.
Analysts were spending 8–12 minutes per encounter just to understand what went wrong before they could even decide what to do next. This created four compounding problems:
Delayed Decision Making - analysts hit decision bottlenecks that cost time and delayed recoverable revenue.
Cognitive Overhead - complex data scattered across disconnected components left analysts mentally fatigued.
Missed Filing Deadlines - without a clear priority hierarchy, time-sensitive claims slipped past appeal windows and became write-offs.
Low Confidence - without clear guidance, analysts second-guessed their actions, re-running the same analysis in circles.
The result: delayed corrections on denied claims translated directly into millions of dollars of recoverable revenue lost per quarter.
Ideate - User Persona
Designing for three distinct operational perspectives
To ensure Next Best Action delivered tangible value across the organization, I mapped the ecosystem into three archetypes -
Connecting high-level financial strategy down to daily desk-level execution:
Executive Leadership (VP of RCM): Needs predictable cash flow and instant visibility into high-dollar denial trends to prevent revenue leakage before appeal windows close.
Operations Management (General Manager): Focused on floor throughput, SLA compliance, and eliminating bottlenecks by routing the right denials to the right specialists automatically.
Core End User (Senior Specialist & Team Lead): Drowning in fragmented payer portals and complex clinical charts; needs instant root-cause clarity and a single, verified next action without cognitive burnout.
Ideation – Information Architecture
Structuring the journey from intake to actionable decision
To transform dense, multi-source claim data into immediate decisions, the information architecture was structured into a linear, distraction-free journey. Instead of overwhelming analysts with an open-ended dashboard, the flow guides them through four focused stages:
Encounter Intake & Lookup:
A clean, focused entry point where analysts input single or batch encounter IDs, supported by quick-search tags and instant validation.AI Data Extraction & Synthesis:
A transparent processing stage that aggregates EHR records, payer rules, and denial history in seconds, setting clear expectations before results appear.Action Directive Workspace:
The primary decision hub. It elevates the most critical recommendation first -
featuring the Next Best Action directive, confidence score, filing deadline, and an explicit 3-step execution plan.
Diagnostic Evidence Matrix:
On-demand diagnostic layers that allow analysts to validate the AI’s reasoning without cluttering the primary view - including root cause breakdown, clinical documentation gaps, and historical appeal patterns.
Final Designs
Refining the dashboard experience
The final design turns a complex denial-management workflow into one question: what should I do next? Users enter an Encounter ID, Patient ID, or Claim number, and within seconds the system pulls the data and surfaces one recommended action.
At the top, the Next Best Action card shows the action, its priority, and the deadline, so users know what to do before reading anything else. Below it, step-by-step instructions (Investigate, Draft Appeal, Prepare for Submission) guide them through the work. Supporting analysis for root cause, clinical data, required documents, success likelihood, and pattern recommendations sits in modular cards that can be scanned at a glance or expanded for detail.
Color-coded statuses and visible success rates show users why the system recommends an action, which builds trust. Light and dark themes, a persistent AI Assistant, and a last-sync indicator make the tool comfortable for long working sessions.
Everything I wrote comes from the screens you shared, not from your research or process, so I can't fill in things like your role, timeline, team, or results. Send me those details and I'll write the other sections to match.
