About GaiaLab

Biological reasoning that shows its work.

GaiaLab is an independent biological-intelligence research platform that turns gene panels and scientific questions into cited, testable hypotheses. It combines evidence retrieval, mechanistic reasoning, ranking, provenance, and prospective evaluation — while preserving the assumptions, timestamps, outcomes, and failures needed to challenge its conclusions.

GaiaLab is building a global ledger of biological evidence: not a database that merely returns associations, but a reasoning and accountability layer connecting evidence, hypotheses, predictions, timestamps, experiments, and outcomes.

Independent research Research use only Prospectively tested Negative results preserved

Mission

Biology does not lack data. It lacks trustworthy synthesis across fragmented evidence. Public biological knowledge is spread across genomic resources, protein databases, pathway systems, disease associations, drug databases, clinical-trial registries, literature and experimental datasets — each internally coherent, none designed to answer a research question end to end.

GaiaLab exists between those sources and the research decision. It is built to help a researcher move from a gene panel to evidence, to mechanisms, to contradictions, to candidate interventions, to testable hypotheses, and finally to prospective evaluation — without hiding uncertainty or pretending computational evidence is experimental proof.

What GaiaLab does

Evidence-backed reasoning

An analysis can draw from 76+ integrated biological data sources, covering evidence synthesis, pathway enrichment, mechanism reasoning, contradiction surfacing, drug and intervention prioritisation, and per-claim source provenance. The active source set varies with credentials, source availability, query context and provider health — so GaiaLab reports which sources actually answered, including when it ran on partial data.

Testable hypotheses

The output is a hypothesis you can argue with: narrated reasoning, ranked mechanisms, surfaced contradictions and a proposed next experiment. It is designed to be challenged experimentally, not to serve as an authoritative medical answer.

Prospective accountability

Important studies freeze the question, the evidence boundary, the prediction or scoring rule, the comparator, the endpoint and the decision rule before the outcome is known. Negative results stay visible: a prospective protocol that only publishes favourable outcomes is not a prospective protocol.

Proof-of-anteriority

Eligible prediction commitments are hashed with SHA-256, batched into Merkle commitments and submitted through OpenTimestamps. Cryptographic verification can establish that a committed prediction package existed no later than a verified Bitcoin block. It establishes timing and integrity — not whether the scientific prediction is correct. Only records with a confirmed receipt are described as Bitcoin-confirmed; /verify shows each record's exact proof status.

How to read GaiaLab evidence

Not all evidence on this site carries the same weight. Three levels, kept strictly apart:

Current scientific status

Drug repurposing
retrospective
RETROSPECTIVE Retrospective benchmarks show measurable but limited correspondence with existing clinical evidence. Live metrics, confidence intervals and caveats are published on /calibration. These are retrospective benchmarks — not clinical validation and not forward-prediction accuracy. A small temporal holdout has also been studied, but GaiaLab treats it as a weak retrospective proxy rather than prospective accuracy.
Phase 9 — DepMap 26Q3
prospective, complete
NOT SUPPORTED The frozen GaiaLab intervention-confidence score did not predict prospective 26Q3 gene-dependency rankings better than the frozen 26Q1 historical-dependency baseline under the predeclared Phase 9 protocol. ρ=0.140 vs 0.729; Δρ=-0.589; 24 of 48 planned pairs were scoreable.

This means one frozen prospective hypothesis was not supported. It does not mean GaiaLab as a whole is disproven, that drug repurposing has failed, that GaiaLab has received independent validation, or that any therapeutic has clinical validation. We report this negative result because the accountability mechanism worked as designed.

Independent post-release integrity review: PENDING. No AI system, CI pipeline or the founder is the independent reviewer.
Drug-repurposing lockbox
prospective, sealed
SEALED — READOUT 2028 A separate pre-registered drug-repurposing study remains sealed, with readout in 2028. It is distinct from the Phase 9 DepMap experiment and has not been validated. Protocol and schedule: /validation.
LassaAI
prospective, awaiting first score
SKILL NOT YET ESTABLISHED The LassaAI v1.1 prospective protocol and freeze are in place. Its prospective skill has not yet been established, and no forecast has been scored. Retrospective Lassa performance is not prospective validation. Separately, a pre-existing NCDC surveillance baseline has been recorded to establish what was already visible when monitoring began — those entries are a baseline boundary, not prospective first observations. More: /lassa.
Lassa host-target evidence
literature retrieval, expert review pending
NOT INDEPENDENTLY VALIDATED On 12 October 2026, GaiaLab completed a reproducible PubMed evidence-retrieval run examining AXL and MBTPS1, with DAG1 and LARGE1 retained as mechanistic context controls. A separate, earlier GSPT1/CC-90009 record is a known-literature positive-control recovery, not a new discovery; it was not rerun in this October workflow. The run passed engineering and evidence-structure safeguards, but retrieved papers, compound-target hypotheses, contradictions, and safety liabilities still require claim-level independent scientific assessment. No Lassa antiviral discovery, efficacy, clinical suitability or endorsement is claimed. This host-target research is separate from the prospective LassaAI forecasting study.

Founder & research governance

GaiaLab was founded and is independently developed by Oluwafemi Scufield Idiakhoa, an Independent AI Researcher. He is the accountable human researcher for GaiaLab's scientific governance, source provenance, study freezes, release decisions and public claims.

His experience surviving Lassa fever informs GaiaLab's attention to neglected infectious diseases. That lived experience motivates the research direction; it is not evidence of model performance or clinical effectiveness.

He holds the BlueDot Impact Professional Certification in Biosecurity, which informs GaiaLab's approach to biological risk, dual-use awareness and responsible AI in biotechnology. It is not a biology degree, clinical qualification, regulatory approval, institutional endorsement or independent scientific validation.

Independence, conflicts & AI assistance

Funding
No external institutional or grant funding. GaiaLab is independently developed.
Conflict of interest
The founder owns GaiaLab and has a financial interest in the platform.
Licensing
LassaAI is open source under the MIT License. GaiaLab's core is proprietary.
AI assistance
Generative-AI systems are used for engineering assistance, evidence organization, consistency checking, and drafting/editing. They are not treated as authors, clinicians, biologists, peer reviewers, or independent scientific authorities. Final scientific claims and releases remain the responsibility of the accountable human researcher.
Endorsements
No claim of endorsement by any agency, institution or expert.

Limitations

Research use only. Do not upload protected health information, direct identifiers or identifiable patient records. All outputs require independent expert and experimental validation before any clinical or therapeutic use. GaiaLab claims no regulatory approval or compliance certification.

Research artifacts

Contact

Research & partnerships: partnerships@gailabai.com