Livemore Bio documentation

Livemore Bio is the computational-biology platform of Livemore Health & Biosciences. It is built for biotech and pharmaceutical research: to represent a specific human's biology, test an intervention against it, and say exactly how far each answer can be trusted.

Research use only

Livemore Bio is research infrastructure. It is not a medical device, not clinical decision support and not dosing guidance. Nothing it produces is a measurement taken from a person, and every result says so. The six reference programs are 0 of 6 validated in vivo. The physical LIFE Patch and the Cendos wet lab are not built yet.

These pages describe release stable 6588707 (commit 658870775e581c044a8832fd107e4097f7a30959), published at github.com/MannyAmah/livemorebio-dist.

The mission

Livemore Bio is being built toward one target: an individualized, living, biological replica of a human being. That is not an avatar or a dashboard. The goal is a digital human whose organs, tissues, cells, genes, proteins and molecules do their biological work and respond according to that person's own properties, so that Person A's twin is not Person B's.

Three connected systems carry that target:

Aegle, the digital twin. The aim is one individualized twin per person, from atoms and molecules up to whole-body regulation.
LIFE, the measurement plane. It is designed around the LIFE Patch, a purpose-built skin-worn sensing device (specified, not yet manufactured). The physical patch will belong to the real human and its virtual counterpart to that human's twin, with both using the same device contract.
Cendos, the lab. It runs computational protocols today and is planned to include a real physical wet lab, so that predictions can be checked against independent measurements.

The systems are designed to talk in both directions. One rule already applies throughout: results, perturbations and observations are never invented to make a demonstration succeed.

What this release is

This release is the working foundation for that mission, not the finished replica.

Area	Status in this release
Body systems in the whole-human contract	15 of 15 registered and reference-backed
Body systems with executable, coupled models	2: endocrine (glucose–insulin) and cardiovascular (ventricular electrophysiology). Their coupling rule is a labelled hypothesis
Genome-scale human metabolism	Human-GEM 2.0: 12,931 reactions, 8,461 metabolites, 2,848 genes, run in Cendos
Reference programs	6. Each one runs, runs partially, or refuses and names the missing measurement
Person-calibrated, context-validated or independently replicated systems	0
Physical LIFE Patch hardware	B0 EVT design target. Not manufactured
Cendos physical wet lab	Planned. Not built

Who it is for

Pharmaceutical and biotech R&D teams who want mechanistic, individualized hypothesis testing with a full audit trail.
Translational and computational scientists who need executable physiology, genome-scale metabolism and reproducible notebooks in one place.
Research, device and laboratory partners who can supply the physical measurements that turn a model into evidence.
Technical evaluators doing due diligence. See how to evaluate Livemore Bio.

What makes it different

Principle	What it means in practice
Individual by construction	Each twin keeps its own parameters and their sources. Cohorts are frozen sets of distinct synthetic member profiles, not one template with noise added to the outcome.
Mechanistic and interpretable	The engines are published models where they exist: Bergman glucose–insulin ODEs, the Hovorka insulin model, O'Hara–Rudy CiPA 2017 electrophysiology and Human-GEM. In-house or heuristic components, such as metformin's pharmacodynamic gains and the metabolic-to-cardiac coupling, are labelled as research assumptions.
Refuses instead of inventing	Ask a question the evidence cannot answer and you get a typed refusal that names the missing measurement, not a plausible-looking number.
Evidence travels with every number	Every result carries its evidence and qualification state and its controls. Depending on the protocol, it also carries applicability, allowed and prohibited claims, and the pinned source and solver behind it.
Infrastructure, not a web app	Three services run from the terminal. The browser console, Python SDK, REST API and Jupyter all see the same twin, cohort, device and evidence identities.

Where to go next

Install Livemore Bio on macOS, Linux, WSL2 or Docker.
Run the quickstart to start the three services and open the console.
Read the evidence model to understand what every result does and does not claim.
See the roadmap for what is built and what comes next.