INTELLIGENCE / Healthcare AI
AI Beyond Automation
Automating a bad process only produces bad decisions faster.
Core concept
The Intelligence Is in the Filters
AI value comes from identifying:
- 01Which cases need intervention
- 02Which cases should pass through
- 03Which providers show outlier behavior
- 04Which clinical factors modify risk
- 05Which utilization patterns signal inappropriate care
- 06Which interventions can materially change cost or outcomes
AI filter model
Ten filters, two pathways
Every case passes 10 filters before anyone decides how much attention it deserves.
All cases
Ten filters
- Clinical factors
- Risk
- Utilization history
- Provider behavior
- Patient factors
- Site / setting
- Prior history
- Cost signals
- Pattern anomalies
- Evidence-based criteria
Low-risk / appropriate
STREAMLINE
- Approve
- Automate
- Reduce friction
Intervention candidates
INTELLIGENT REVIEW
- Clinical decision support
- Human review
- Provider intervention
- Alternative pathway
- Additional information
Where this was applied
Clinical, financial, utilization, quality and longitudinal-care filters
AI should not simply automate existing healthcare workflows. Its greater value is in applying clinical, financial, utilization, quality, and longitudinal-care filters that allow the system to make more intelligent decisions.
Clinical, financial, quality & SDOH risk
Use AI and EMR data to identify clinical, financial, quality and SDOH risks—not just generate individual alerts.
- Pieces Technologies
AI + EMR clinical and healthcare-performance strategy
BuiltAdvised on product strategy connecting AI and EMR data to clinical, financial, quality and SDOH risks.
ImpactProduct strategy addressed STAR measures, HEDIS, Total Performance Score, ACO financial performance, readmissions, length of stay, hospital-acquired conditions, sepsis, uncompensated care and social determinants of health.
Longitudinal care & authorization
Use pathway intelligence to anticipate future care and appropriate downstream authorization, rather than automate isolated review transactions.
- New Century Oncology / Evolent Health
Longitudinal oncology & cardiology pathway decision support
BuiltDesigned a new Oncology and Cardiology decision-support system based on care pathways.
ImpactThe design allows clinically appropriate downstream requests to be automatically approved as a patient proceeds through the expected pathway, rather than repeatedly treating care as unrelated transactions.
Specialty-program strategy & pathway intelligence
Initial founding-team consultation in 2019 led to MSK development, followed by Cardiology consulting and later AI-supported pathway intelligence, sleep and genetic-testing algorithms, and specialty-drug programs.
- Cohere Health
MSK clinical-program development strategy; Cardiology clinical-program development strategy; AI-supported specialty pathway intelligence; AI-driven sleep-testing & genetic-testing algorithms; Targeted therapy, immunotherapy & monoclonal-antibody management
BuiltConsulted with the initial founding team — CEO, COO and product lead — in 2019, leading to development of Cohere Health’s MSK program.
ImpactThe initial founding-team consultation in 2019 led to development of Cohere Health’s MSK program. Further consulting supported Cardiology development and a next-generation utilization-management model focused on clinical pathways, specialty programs and selective intervention.
Clinical utilization & service architecture
Connect clinical, utilization and longitudinal-care filters to specialty programs, peer review, care management and the service model.
- Hinduja Healthcare Services → Sagility
BPaaS / UM-as-a-Service architecture
BuiltCreated overall UM strategy covering prior authorization, peer review, clinical decision support and care management.
ImpactThe work was part of preparing the healthcare business for acquisition by Baring Private Equity Asia / BPEA.
- HealthHelp
Collaborative specialty utilization-management programs; Integrative Oncology; Protos clinical rules & authorization engine; Spend-reduction products: Site, Scope, Triage, Direct, Concierge & Access; Utilization-reduction products: Consult, Access, Inform, Guide & Alert; Provider Score, savings valuation & risk / predictive analytics
BuiltRestructured product strategy, health informatics and client-service delivery.
ImpactRevenue grew from approximately $35M to more than $60M during executive tenure, with approximately $5M EBITDA growth and more than $20M revenue associated with new programs and enhancements.
Comparison
Two ways to use the same technology
Automation-first AI
- Processes more cases
- Reduces manual work
- Speeds existing workflows
- Measures productivity
- Applies automation broadly
- Creates administrative efficiency
Intelligence-first AI
- Determines which cases matter
- Identifies intervention opportunities
- Reduces unnecessary intervention
- Supports better clinical decisions
- Applies intelligence selectively
- Targets clinical and economic value
If AI simply automates the existing workflow, its principal benefit may be administrative efficiency. To bend healthcare cost or improve clinical outcomes, AI must identify the cases where a different action changes the result.
Next step
Find where AI changes the clinical or economic result.
Bring the workflow you are automating. We will ask which cases actually need a different action.