Personalized Medicine, Impersonally Delivered by AI
Since the start of the Human Genome Project, we’ve pursued tailored treatments with relentless hope. Yet, as rare diseases rise, care delivery has become increasingly automated, impersonal, industrialized, and soon will be systematically AI-enabled.
Now, we sit at the intersection of AI, genomics, and healthcare economics. My own journey, from statistical genetics in academia to leading AI strategy at a CDMO, has shown me just how disconnected our ambitions can be from the infrastructure we’ve built. We’re chasing personalized care through systems that are anything but personal.
The question we must confront is: What kind of infrastructure do we need to truly deliver personalized medicine — and how do we scale it without losing the personal touch our patients need?
The Problem: Misaligned Infrastructure
Personalized medicine is not failing because we lack insight, despite certain limitations in the science. We know how to sequence genomes. We understand how certain pathways could affect the response to therapy. And we are beginning to understand the molecular basis of ultra-rare conditions.
To take a few steps ahead, what we lack is an infrastructure that can translate insight into personalized care.
Today, much of our technical infrastructure, from electronic health record (EHR) systems to AI-based tools, prioritizes scalability, speed, and standardization. These traits may improve efficiency, but they often come at the expense of nuance, empathy, and clinical context.
In an industry that touches human life so directly, this trade-off is costly.
Three Layers of the Solution
The solution lies in rethinking the very foundation of how personalized care is delivered, structured across three interconnected layers:
1. Context-Aware Data Systems
It’s not enough to have interoperable systems. We need interpretable, context-rich data that captures the full complexity of human life: genomic variations, phenotypic presentations, social determinants of health, and real-world outcomes.
AI must learn from the edges of the distribution, not just the statistical mean.
Get Mo Villagran’s stories in your inbox
Join Medium for free to get updates from this writer.
2. Governance with Purpose and Care
Governance must go beyond compliance. We need frameworks that ensure fairness, transparency, and accountability in AI-assisted decisions. This includes expert-in-the-loop design, patient consent frameworks, and robust protocols for bias detection.
The system cannot scale if it undermines public trust.
3. Patient-Centered Design in Care Delivery Systems
Whether the interaction happens through an app, a voice assistant, or a clinician-supported workflow, the care delivery system must preserve human dignity. We must design for agency, empathy, and clinical intuition, not just throughput.
The goal isn’t just to make AI smarter — it’s to make it care about humans.
Where We Go From Here
The infrastructure of personalized medicine must evolve — from something built to serve systems to something designed to serve patients.
This will not be an easy journey.
It requires uncomfortable conversations across disciplines.
It requires executive alignment on long-term value, not just short-term ROI.
It demands that technologists, clinicians, payers, and numerous key stakeholders start working in sync.
But change is possible. We can build a better system together.
Aurelian Advisory — Strategic Insights at the Intersection of AI, Alignment, and Execution.
Where Your Vision Finds Its Voice. 🌐 https://aurelianadvisory.com
If you’re a healthcare leader, innovator, or policymaker wrestling with this tension, I’m ready to listen.
Mo Villagran, Executive Advisor 📩 mo@aurelianadvisory.com
