Superpower Daily
Anthropic Nears $15 Billion Credit Facility Before Its Public I...GPT-6 Astra Hits 99.9% on ARC-AGI-3, but Scores 62.7% in a Shar...Crusoe Signs Reported $13 Billion, Five-Year AI Cloud Deal With...Tesla Says 45 Cybercabs Are Registered for Texas Robotaxi Netwo...Microsoft Releases MAI-Transcribe-2 at 10 Cents an HourGE Appliances Deploys 800 AI Agents, Reports 25% Fewer Back Ord...Anthropic Nears $15 Billion Credit Facility Before Its Public I...GPT-6 Astra Hits 99.9% on ARC-AGI-3, but Scores 62.7% in a Shar...Crusoe Signs Reported $13 Billion, Five-Year AI Cloud Deal With...Tesla Says 45 Cybercabs Are Registered for Texas Robotaxi Netwo...Microsoft Releases MAI-Transcribe-2 at 10 Cents an HourGE Appliances Deploys 800 AI Agents, Reports 25% Fewer Back Ord...
Businesspublished

Gelomics Teams With Google Australia to Build AI for Cancer Trial Design

The biotech’s proposed system combines patient-derived tumour models with molecular profiling to predict drug response before trials. Its first stated use is trial design, not treatment decisions for hospital patients.

By 2 min read
Gelomics Teams With Google Australia to Build AI for Cancer Trial Design
Gelomics Teams With Google Australia to Build AI for Cancer Trial Design

Listen to this story

The audio brief

About 1:40
0:001:40
Read transcript
Gelomics and Google Australia are building an AI platform to predict which patients may respond to an experimental cancer drug before a clinical trial is designed. The partners are targeting a prototype for early-stage drug development by mid-2027, but they have not reported trial results or clinical performance yet. The system begins with LunaX, Gelomics’s method for growing three-dimensional tumour models from patient biopsies inside a specialised gel. Unlike a flat plastic dish, the model is intended to better reflect how cells behave in the body. Gelomics says these models can be established in two to three weeks and kept for longer-term testing. Google Cloud’s Gemini Enterprise Agent Platform is being used to optimise the conditions for growing those tissues. The goal is greater standardisation, so researchers can more clearly separate a drug’s effect from variation in the biological sample. In parallel, Gelomics and the Queensland Spatial Biology Centre are mapping proteins and genes cell by cell in the original tumour. Machine-learning models will then connect those molecular maps with laboratory drug responses and clinical outcomes. If that link proves reliable, sponsors could design trials around people most likely to benefit, potentially making them smaller and reducing exposure to ineffective treatments. Those are proposed benefits, not demonstrated outcomes. The important boundary is that this is trial-selection technology first. Using it to recommend treatment for an individual hospital patient would require clinical validation. The key question through 2027 is whether the models can predict response consistently enough to influence real drug-development decisions.

Story brief

3 key points

Gelomics and Google Australia are developing an AI-assisted platform to make cancer-drug trials more selective, using patient-derived 3D tissue models and tumour molecular maps to identify likely responders. The near-term product is for early-stage drug development—not treatment decisions for individual patients, which still requires clinical validation. Gelomics plans a prototype by mid-2027. If the modelling...

  1. 01

    LunaX grows biopsy-derived 3D tumour models in specialised gel and can retain them for long-term testing.

  2. 02

    Google Cloud’s Gemini Enterprise Agent Platform helps optimise cell-culture conditions and improve model standardisation.

  3. 03

    Gelomics and the Queensland Spatial Biology Centre are mapping tumour proteins and genes cell by cell.

Gelomics has partnered with Google Australia on a cancer drug-discovery platform meant to identify which patient groups may respond to an experimental treatment. The company is targeting clinical-trial design first, while individual treatment guidance remains a longer-term goal requiring clinical validation.

A tissue model instead of a flat dish

The platform starts with LunaX, which grows three-dimensional human tissue models in a specialised gel, including tumours derived from patient biopsies. Google characterizes the approach as an alternative to conventional flat plastic dishes, which it says cannot reproduce how cells behave inside the body.

Gelomics says LunaX can establish standardised tissue models within two to three weeks and retain them for long-term testing. It uses Google Cloud’s Gemini Enterprise Agent Platform to predict and optimise the cell-culture conditions needed to produce those models; Gelomics says standardisation helps separate a drug effect from variation in the biological sample.

Connecting a biopsy to a trial cohort

Alongside the lab-grown model, Gelomics maps each original biopsy cell by cell with the Queensland Spatial Biology Centre. That produces molecular data on the arrangement of proteins and genes in a tumour. The company then trains machine-learning models to link those maps with laboratory-tested drug responses and clinical outcomes, with the goal of predicting drug efficacy from a molecular map alone.

Google and Gelomics position that prediction step as a way to design smaller, more targeted trials around patients most likely to benefit. They say this could reduce exposure to ineffective therapies, avoid losing viable drugs tested in unsuitable patient populations, and lessen reliance on animal testing. Those are intended benefits rather than reported trial outcomes.

Gelomics aims to release a prototype for early-stage drug development in mid-2027. The partnership builds on the company’s involvement in Google’s AI First Accelerator Program, making the next milestone a drug-development prototype rather than a system for clinical care.

Sources

  1. blog.googleHow Gelomics is Redefining Cancer Drug Discovery

Related coverage

Continue reading