Here’s Part II of our deep dive into the state of AI in healthcare. Our last “AI 101″ post introduced the concept of the four-stage AI adoption maturity lifecycle. So, what’s next?
It’s artificial general intelligence (AGI), but is it reflected in the AI maturity lifecycle (pictured below)? If not, why not? Let’s answer that question by digging into AGI to find out what it means to you and your organization.
AGI, also known as “strong AI,” is a hypothetical form of AI that possesses human-like intelligence and can perform any intellectual task that a human can. Unlike today’s AI, AGI would be capable of understanding and reasoning across a broad range of tasks. It would not only replicate or predict human behavior but also embody the ability to learn and reason across diverse scenarios, from creative endeavors to complex problem-solving. To do that, it would require not just intelligence but also emotional and contextual awareness.
This type of intelligence could potentially manage diverse and complex tasks that require creativity, emotional intelligence, and multi-dimensional thinking—capabilities far beyond the reach of today’s AI.
Hypothetical Attributes of AGI:
- General purpose: Not limited to specific tasks like image recognition or language translation but would be capable of performing any intellectual task that a human can, according to Gartner
- Human-level intelligence: Has the ability to reason, learn, solve problems, and understand the world in a way that is comparable to human intelligence
- Autonomous learning and adaptation: Able to learn from data and experience, improve its performance over time, and adapt to new situations without explicit programming
- Abstract thinking and problem-solving: Capable of abstract thinking, understanding complex concepts, and solving problems in novel ways
- Common sense and reasoning: Possesses common sense reasoning abilities, allowing it to understand the world and make decisions in a way that is similar to humans
However, the journey toward AGI is constrained by our current understanding and technological limitations. Building machines that truly understand and interact with the world like humans involves not just technical advancements in how machines learn, but also profound insights into the nature of human intelligence. Current AI lacks the ability to fully comprehend context or develop consciousness, which are critical for tasks that humans navigate seamlessly.
As AI evolves, recognizing the profound distinctions between AI and AGI is imperative. While AI already improves our daily lives and workflows through automation and optimization, the emergence of AGI would be a transformative leap, radically expanding the capabilities of machines and redefining what it means to be human—with far-reaching implications and impacts for society, employment, and the future of our world. Given this reality, AGI may well evolve into stage five on the adoption maturity lifecycle more rapidly than we think, so awareness is essential.
“AI will deliver value by unburdening clinicians and empowering them to focus on caring for their patients… Trust is key to the adoption of AI. AI will be deployed at the speed of trust!”
– Stephen Jones, MD, CEO of Inova and Immediate Past Chair of the Board, AMGA, at the Closing Keynote Panel at the 2025 AMGA Conference
________________________________________________________________
Example Application: Population Health Management
Goal: Improve Clinician and Nurse Productivity/Capacity while Reducing Fatigue & Burnout
Target Personas: Physicians, Nurses, and Care Managers/Care Coordinators
High-Value Use Cases for AI: Population Health and Care Management, Physician-Patient Encounters
Target jobs to be done (JTBD) for AI augmentation and automation:
- Digital front door / call center patient self-service
- Predict risks for 30-60-90-day ED admissions/readmissions in attributed patient population
- Identify high- and rising-risk patients with prescribed care gaps for care management
- Enroll high- and rising-risk patients into a virtual care orchestration program with Deviceless RPM
- Physician-patient encounter, documentation, and follow-ups
Portfolio of AI modalities for augmentation and automation and build/buy/partner decisions:
- Machine learning and deep learning [in-house decision science team]
- Gen AI/LLMs [in-house decision science team with GenAI/LLMs]
- NLP/NLG [in-house decision science team with support from offshore SI Consultants on contract, as necessary]
- Agentic AI/ AI agents [Buy Platform that aligns with operating priorities]
So What? Business Case (Value/ROI/Payback) for AI augmentation and automation: (All $$ values are approximate)
- Increase cost-effectiveness of customer experience/call center solution – no additional FTEs in current fiscal year ($ 675 K)
- Enhance clinician, nurse and care manager productivity ($ 550 K)
- Improve nurse and caregiver capacity – no additional FTEs in current fiscal year ($ 1.3 M)
- Increase savings from avoidable ED admissions and readmissions in 2025-26 ($ 5.75 MM)
- Maximize value-based care (VBC) reimbursements – ($ 4.5 M in 2025)
- Lower clinician, nurse and care manager fatigue and burnout and improve retention ($ 3.75 M)
Lightbeam AI has achieved an average 41% relative reduction in avoidable admissions and 23.6% decrease in readmission risk.
Combining Lightbeam Advisory Services, a VBC Lightbeam client achieved a 39.45% decrease in per-member-per-month costs for a high-risk patient cohort, resulting in more than $23 million in savings. Read more.
Total Value from AI innovation in population health and care management in 2025: $10.77 M
Total AI investment in this initiative(s): $ 3.5 M
Anticipated ROI on AI for this initiative(s): 307% or 3X
Payback Period: 5 months (estimated)
Applying the AI innovation strategy and 4 stage AI adoption lifecycle to population health management and care management.
From strategy-to-execution—how to lead with value and ROI driven AI innovation aligned with the ai adoption maturity life cycle.
The 3X3 Matrix in figure 4 is a valuable ideation and planning tool to morph the AI innovation strategy in figure 3 and the previous section into specific use cases that can be augmented or automated with AI on the X-axis, based on the anticipated value and ROI on the Y-Axis (modest/ low, significant/medium or game-changing/high) as shown above.
This is a useful tool to determine the level of AI enablement that is aligned with the use cases/job-to-be-done and the target user’s needs to ensure that the appropriate AI modality is deployed without “over kill” (“a jack hammer for every nail”). For instance, agentic AI/ AI agents to automate the underlying task, process or workflow may not be needed when a healthcare organization is looking for prescriptive actioning and recommendations for their care managers, or simply early identification / prediction of high-risk and rising-risk patients.
We will leverage this 3X3 matrix to articulate the most valuable use cases for care managers and case managers as well as clinicians in population health management here:
AI augmentation
Use Case: Predicting high-risk and rising-risk patients from your attributed population for care management/care coordination.
Value (Modest): Lower risk of penalties, fines, and losses from 30-60-90-day re-admissions/ extended length of stay.
AI prescriptive actioning
Lightbeam Health Solutions Use Case: Reduced ED Visits by 7% by prescribing actionable care gaps for high-risk and rising-risk patients for proactive care orchestration.
Value (Significant): Incremental cost savings (avoidable ed visits) from preventing rising-risk patients morphing to high-risk patients through virtual care management (rpm) and higher VBC reimbursements.
Lightbeam has recently achieved impressive results for Saint Peter’s University Hospital in New Brunswick, New Jersey. Using Lightbeam AI to integrate SDOH data, Saint Peter’s cut ED visits by 7% by precisely identifying and prioritizing rising-risk patents for intervention and automating workflows. By applying key social factors such as food insecurity, lack of transportation, and lack of personal support, the care team efficiently matched highest-need patients to appropriate social programs to significantly improved outcomes. Key highlights included:
- 7.1% reduction in ED visits among high-risk patients
- 9.5% ED visit rate for patients receiving AI-guided interventions vs. 16.7% for similar patients without intervention
- Increased use of food and transportation support programs
Lightbeam AI enabled Saint Peter’ University Hospital to proactively address hidden social needs, improve care for vulnerable patients, and deliver measurable results. Read the detailed case study here.
AI automation
Use Case 1: Digital front door conversational AI agent for 24×7 patient self-service providing access to healthcare information appointment scheduling, refills, eligibility, and bill payments
Value (Significant): Reduced customer experience/call center and administrative costs like FTEs, reducing wait times for patients and higher customer satisfaction
Use Case 2: IVR based conversational/voice AI agents integrated with RPM to engage with and enroll new patients into care management program, or scheduling appointments with PCPs
Value (significant): Savings in nurse care manager/ coordinator or call center agent FTEs ($$$)
Use Case 3: Clinical AI agents with IVR to help clinicians verbally query EHRs, extract patient chart summaries, document notes and automate follow-up actions, such as lab tests or referrals.
Value (Game-Changing/Exponential): Improve physician productivity (higher # of patients/day) while reducing administrative tasks, fatigue, and burnout.
Summary
AI investments, innovation, deployment, and adoption have been marked by disproportionate hype, incessant noise, confusion, and an unfortunate absence of value and ROI from the billions invested thus far.
Most of the venture capital spend, as well as media hype on AI, has been disproportionately focused on GenAI/LLMs; however, Gen AI/LLMs has not always lived up to the fervor and have not delivered on the promise.
Only 43% of healthcare organizations who have deployed AI applications have seen measurable value and ROI from their investments which is indeed sobering.
Follow Lightbeam AI to learn more about empowering your team and organization to deliver high-quality, cost-effective care by combining advanced analytics, prescriptive insights, and scalable solutions tailored to the complexities of modern healthcare systems. And watch our first LinkedIn Live: Part I – The Art of the Possible with AI in Healthcare.
Please feel free to share this detailed blog post with your colleagues to inform your AI strategy in 2025 and beyond.
As always, I welcome your comments and feedback on my personal LinkedIn blog or via email at ade@lightbeamhealth.com at Lightbeam Health Solutions.
