Before you treat the body,
simulate it.
The premise of the digital twin is almost shockingly simple: build a computational model of a specific patient — their genome, their cellular architecture, their metabolic rhythms, their immune landscape — and use it as a sandbox. Run trials on the model, not the person. Fail safely in silicon.
In engineering, digital twins have existed for decades. NASA runs virtual copies of spacecraft. Aircraft manufacturers simulate every stress condition before a plane ever flies. The principle is identical in medicine, but the subject is incomparably more complex. A human being isn't a turbine — it's 37 trillion cells operating in parallel, each one a living computer, all of them talking at once.
AI is the bridge that makes this possible. Not because AI understands biology — it doesn't, not yet — but because AI can find patterns in complexity that human cognition cannot. A model trained on millions of patient outcomes, cellular datasets, and molecular interactions can simulate how a specific body will respond to a specific treatment with a fidelity no individual doctor could replicate from memory alone.
"Digital twins create virtual copies of patients that help predict health issues, simulate treatments, and continuously monitor health in real time."
— Journal of Biomedical and Health Informatics, 2025The result isn't just better medicine. It's a fundamentally different relationship between a patient and their disease — one where the future is partially observable before it happens, where a treatment failure can be discovered virtually rather than physically, and where personalised care becomes something more than a marketing phrase.
What a digital twin
is actually made of.
Each layer is a different resolution of the same patient. The genomic layer captures what is possible; the clinical layer captures what has happened; the layers in between capture what is currently happening. A true digital twin integrates all six — and the AI's job is to hold them in coherent relationship to one another, updating in real time as the patient's condition changes.
This is where most current systems fall short. Integrating multi-omics data — genomics, transcriptomics, proteomics, and metabolomics simultaneously — remains one of the hardest open problems in computational biology. The data layers don't always agree. Resolving those contradictions is where the interesting science lives.
What digital twins
can already do.
Drug Target Discovery
AI turns genes on and off in a virtual cell model to identify which ones are implicated in disease — narrowing thousands of candidates to a handful worth testing in the lab.
Active NowVirtual Drug Trials
Instead of synthesising thousands of compounds, AI screens millions computationally — reducing wet-lab work from years to months and arriving at potent candidates with far fewer failures.
Active NowPersonalised Dosing
A twin built from a patient's metabolic profile can predict how quickly they'll metabolise a drug — eliminating the trial-and-error that still governs most dosing decisions today.
Active NowNeurodegenerative Prediction
Simulating protein accumulation dynamics years before symptom onset — identifying the tipping point before it's crossed and creating a window for early intervention.
EmergingTumour Microenvironment Modelling
Simulating not just the tumour but its surrounding ecosystem — immune cells, stromal tissue, blood supply — to predict which immunotherapy approach will penetrate and persist.
EmergingWhole-Body Digital Patient
A continuously-updated computational model of an individual that reflects their real-time health state — a living medical record that predicts rather than just documents.
Near FutureRun a virtual drug trial
on a digital twin.
Select a treatment candidate and run it against a virtual patient model. The twin simulates efficacy, toxicity, and resistance probability before a single molecule is synthesised.
// Virtual Clinical Trial — Oncology Model
Select a candidate to beginThe Bivon as
proto-twin.
Bivology has been building digital twins from first principles
Medical Digital Twin
- Built from a real patient's multi-omics data
- Simulates known biology under treatment pressure
- Predicts outcomes within existing biological rules
- Calibrated to Earth life — carbon, water, ATP
- Goal: optimise treatment for a specific person
- Constrained by what biology permits
Bivon / Bivology Twin
- Built from algorithmic rules — code-based DNA
- Simulates novel biology under arbitrary conditions
- Predicts emergent behaviour from first principles
- Unconstrained by carbon chemistry or Earth physics
- Goal: understand what rules produce what life
- Limited only by computational imagination
The medical digital twin asks: given this specific body, what happens next? The Bivon asks something more radical: given these rules, what body becomes possible? Together they bracket the same mystery from opposite ends — the known and the possible, the real and the virtual, the patient and the probe.
This is why Bivology matters as more than a curiosity. The same logic that lets a Bivon evolve resistance to a simulated antibiotic is the same logic, abstracted, that lets a pharmaceutical AI predict resistance evolution in a human tumour. Virtual life isn't a metaphor for real life — it's a laboratory for understanding the rules that real life runs on.
Where the series
goes next.
Signal Networks & Disease Tipping Points
How cell signaling networks collapse from health into pathology — and the AI models learning to predict the exact moment of failure.
Antimicrobial Resistance — Evolution in Fast Forward
Bacteria evolving resistance in real time, simulated as a Bivon population under antibiotic pressure — one of medicine's most urgent challenges.
The Tumour Ecosystem
Cancer as emergent ecology — competing cell populations, immune evasion, and the microenvironment as battlefield. A simulation-first exploration.
The Bivon That Met Its Twin
A Bivology Chronicles story about a digital organism that encounters a perfect copy of itself — and discovers they are no longer the same.