Vishal Patel MD, PhD

What happens to human cognition when we intersect with AI systems designed to minimize cognitive friction? This question drives my current work.

I am building Phronos—an observatory for human cognition in the era of artificial intelligence. The work involves large-scale analysis of human-AI interactions, cognitive phenotyping, and open-source instrumentation.

The premise: to offer tools to fortify ourselves cognitively, so that we engage with AI consciously.

The name comes from phronesis—practical wisdom, the kind that cannot be taught but must be cultivated through observation and reflection.

Member of the Global Wellness Institute's Men's Wellbeing Initiative, working to advance research and awareness around men's physical, mental, and emotional health.

Five years building the scientific foundation for evidence-based luxury wellness. The deepest lesson: prescriptions don't work. Behavior change requires meeting people where they are.

At Sensei, founded by Larry Ellison and Dr. David Agus, I led efforts to transform personal health and wellness information into inspiring experiences to support sustainable lifestyle change.

Over five years, I rose from Director of Wellness Research to Co-President and Chief Medical Officer, and finally to Chief Science and Innovation Officer.

The Sensei Way was built on three paths—Move, Nourish, Rest—and the constant conversation between mind and body that underlies each.

Exploring the intersection of wellness data and longevity science. How do we measure and optimize for healthspan, not just lifespan?

At the intersection of healthcare, machine learning, and clinical operations—building analytic approaches that reduce medical costs and improve quality.

My focus was on the epistemic opacity of the healthcare system: how do we understand and process medical data when the systems generating that data are themselves opaque?

Berkeley-based, mission-driven data science company. Led matrixed teams to provide ML services for payment accuracy, low-value care reduction, and anomalous provider behavior detection for 45+ million health plan members.

Led cross-functional teams designing next-generation data science products. NLP and predictive model development, chart abstraction workflows, and automation of manual clinical tasks.

Here I learned that technical sophistication means nothing if it doesn't fit the workflow. Implementation is its own science.

Founded to organize the exploding volume and variety of wellness information. The paradox of our information age: while information is becoming more readily available, the answers we seek often become more elusive.

Developed predictive models of insurance claims and clinical datasets. Implemented ML algorithms to identify core drivers of clinical outcomes. Served as physician advisor to product development teams.

My academic training established the methodological foundations I continue to apply: the search for hidden patterns in complex biological and behavioral systems.

My doctoral work at Case Western Reserve University employed combinatorial network analysis to identify molecular patterns distinguishing short-term from long-term survivors of glioblastoma.

CRANE (Combinatorially dysRegulated subNEtworks) achieved 80% prediction accuracy using a 50-gene subnetwork signature, demonstrating that network-based mRNA signatures serve as better indicators of protein-level dysregulation than sets of differentially expressed single genes—a nearly 7-fold improvement.

At MetroHealth System and the Louis B. Stokes VA Medical Center, I developed statistical approaches to mine patterns from de-identified patient records—detecting drug side effects and predicting adverse events.

This work showed that EHR data could reveal multivariate pharmacosurveillance signals invisible to traditional case reports.

Using the Explorys platform to analyze population-level electronic health records. Applied NLP algorithms to physicians' notes and collaborated with clinicians to identify clinical endpoints.

As faculty in the Systems Biology and Bioinformatics program at Case Western, I taught graduate courses in Clinical Research Informatics and Healthcare Data Analytics.

Before medicine, I studied engineering. My first love: the quantitative foundations—fluid mechanics, systems dynamics, computational methods—that would later shape how I approach biological and behavioral complexity.

Eight years traversing the molecular foundations of disease while learning to see patients, not just data. The dual degree was about learning to translate between languages: the molecular and the clinical.

My dissertation explored the network architecture of cancer—how patterns of gene expression could predict survival in glioblastoma. This work established my orientation toward network thinking.

A certificate in Ayurveda offered a framework for understanding constitution, imbalance, and restoration that complemented my Western training.

The key lesson: ancient systems encode empirical wisdom that modern science is only beginning to validate.

Bachelor of Science in Mechanical Engineering. Engineers ask: how do we model this system? How do we intervene? These questions never left me.

Wavelet-Based Characterization Of Small-Scale Turbulent Structures In A Mechanical Heart Valve Flow — ASME Summer Bioengineering Conference, 2005 (Winner, Student Poster Competition)

Synthesis and In Vitro Biocompatibility of Injectable Polyurethane Foam Scaffolds — Tissue Engineering, 2006

Looking back, the through-line becomes clear. In glioblastoma, I searched for molecular patterns that could distinguish survivors from non-survivors—and found that the key was not individual gene expression but the configuration of genes in networks.

At Sensei, I sought to transform wellness information into actionable insight—and found that the key was not more data but better conversations.

Now, at Phronos, I'm asking what happens to human cognition when we intersect with AI systems designed to minimize cognitive friction.

The questions have remained consistent: How do we choose the labels by which to define ourselves? What patterns are we missing when we average across populations? What gets lost when we optimize for efficiency?