ProxyBio, a newly launched University of Arizona startup, has secured $3.5 million in funding to scale its AI-powered platform that tests cancer drugs on miniature, patient-derived tumor models, dramatically reducing the time needed to predict treatment effectiveness from six months to six weeks. The company's approach represents a fundamental shift in how researchers screen potential therapies, moving away from traditional methods that destroy tissue samples and offer limited insight into how individual patients will respond to treatment.

Why Are Patient-Derived Tumor Models Better Than Traditional Drug Testing?

Cancer is not a single disease. Each patient's tumor is shaped by unique genetic mutations, tissue characteristics, and cellular interactions. ProxyBio's technology uses living organoids, which are miniature three-dimensional tumor structures grown directly from patient tissue, to test how cancer cells and surrounding healthy tissue will respond to different drugs in their natural environment.

This "living diagnostic" approach stands in stark contrast to traditional preclinical models. Conventional methods often destroy biopsy samples during testing, offering limited insight into treatment effectiveness. ProxyBio's organoids remain alive throughout the screening process, allowing researchers to measure how every cell in a tumor responds to therapy at extraordinary depth.

"Cancer is not one disease. It is thousands of unique diseases shaped by each patient, each tissue, and each cell. ProxyBio's platform is built for that reality. By combining patient-derived organoids, high-content imaging, and AI, we can measure how a living tumor responds to therapy at extraordinary depth," explained Kelvin W. Pond, Assistant Professor and Chief Science Officer at ProxyBio.

Kelvin W. Pond, Assistant Professor and Chief Science Officer, ProxyBio

The significance of this approach lies in what researchers call the "cell neighborhood effect." A cancer cell's behavior depends not just on its own genetics, but on the surrounding cells and their chemical signals. ProxyBio's organoids preserve these critical interactions, providing a far more accurate picture of how a patient's unique tumor ecosystem will respond to a proposed treatment.

How Does ProxyBio's AI Platform Speed Up Drug Testing?

ProxyBio's CLARITYAI platform powers three distinct products designed to accelerate different stages of drug development. The company has already achieved remarkable speed improvements through automation and optimization. What originally took six months to screen a patient's tumor against multiple drug candidates now takes only six weeks.

The company's three-product suite includes:

• ProxyDISCOVERY: Enables pharmaceutical companies to screen potential cancer drugs against patient-derived organoids, identifying the most promising candidates before expensive human clinical trials.
• ProxyTOX: Assesses preclinical safety by testing how drugs affect not just cancer cells, but healthy tissue as well, reducing the risk of unexpected toxicity in human patients.
• ProxyPATIENT: Provides companion-diagnostic-style patient profiling to help clinicians make more informed treatment decisions for individual patients, currently available for research use only as the company seeks FDA approval.

Curtis Thorne, CEO and Associate Professor of Cellular and Molecular Medicine, described the process as "running many clinical trials in parallel for a patient in a dish." By measuring how every cell of a tumor responds to drug treatment in its natural cellular environment, researchers gain insights that traditional methods cannot provide.

Curtis Thorne, CEO and Associate Professor of Cellular and Molecular Medicine

The $3.5 million investment from GKCC will accelerate the company's goal of reducing screening time to just one week. This speed is critical for patients with early-onset colorectal cancer, a primary focus for ProxyBio, where rapid diagnosis and treatment decisions can significantly impact outcomes.

How Can AI Drug Discovery Overcome Data Limitations?

While ProxyBio focuses on patient-level drug response prediction, a parallel challenge in AI-driven drug discovery is the fundamental scarcity of biological training data. The Trillion Gene Atlas project, a collaboration between Basecamp Research and PacBio, is addressing this bottleneck by sequencing genomes from over 100 million previously unstudied species.

Current genomic datasets are heavily skewed toward a tiny fraction of life on Earth. Approximately 70 percent of all public genomic data comes from just five species, creating a severe bias that limits how well AI models can generalize across biology. This data imbalance means most AI models today are good at spotting patterns they have already seen, but struggle when encountering novel proteins, rare organisms, or different environmental conditions.

"Today's genomic datasets are heavily skewed toward a tiny fraction of life on Earth. Imagine if ChatGPT were trained on just five books. As a result, most AI models today don't really understand biology yet. They're good at spotting patterns they've already seen, but they struggle the moment you move into something new, such as a novel protein, a rare organism, or a different environment," noted Glen Gowers, Co-founder and CEO of Basecamp Research.

Glen Gowers, Co-founder and CEO, Basecamp Research

The Trillion Gene Atlas aims to overcome this constraint by generating unprecedented volumes of high-quality genomic data from diverse species. This massive expansion of training data will enable AI models to learn the fundamental "language of DNA" across the full diversity of life, not just the narrow slice currently available in public databases.

What Are the Biggest Barriers to AI Adoption in Drug Development?

Despite promising advances in AI-driven drug discovery, significant adoption barriers remain in clinical development. A poll conducted at the Clinical Trials Technology Congress in London found that trust and regulatory uncertainty are the biggest obstacles, cited by 50 percent of clinical trial professionals surveyed.

However, the outlook is improving. The same poll revealed that 42 percent of respondents are already seeing early signs of return on investment (ROI) from AI in clinical development, with another 23 percent expecting ROI in the near future. Over the next three to five years, clinical development professionals believe AI will have the most impact on data cleaning, data analysis, and insight generation, cited by 48 percent of respondents.

Regulators from the UK's MHRA (Medicines and Healthcare products Regulatory Agency), the Danish Medicines Agency, and the Swedish Medical Products Agency emphasized that they are ready to embrace AI and are eager for pharmaceutical companies to engage early in the development process. The key requirement is that AI approaches be validated, auditable, and explainable, rather than opaque "black-box" models that create uncertainty for both sponsors and regulators.

The poll also highlighted an emerging trend in patient-centric drug development. Sixty percent of respondents are already using, piloting, or exploring patient-generated data, including social media listening, to inform clinical development decisions beyond marketing. More than half, 58 percent, say the primary benefit of social media listening is understanding patient needs, monitoring sentiment, and capturing patient experience.

Steps to Accelerate AI Adoption in Drug Development

• Early Regulatory Engagement: Pharmaceutical companies should initiate conversations with regulatory agencies during the development phase, not after, to ensure AI approaches meet safety and compliance standards before implementation.
• Invest in Model Transparency: Replace black-box AI models with validated, auditable, and explainable approaches that regulators and clinicians can understand and trust for patient safety decisions.
• Expand Training Data Diversity: Support large-scale genomic sequencing projects and biobank initiatives to ensure AI models are trained on diverse biological data, improving generalization across different patient populations and disease types.
• Establish Ethical Data Frameworks: Develop standardized, ethical approaches for collecting and using patient-generated data, including social media insights, to ensure privacy and compliance while capturing real-world patient perspectives.

ProxyBio's rapid progress demonstrates the potential of AI-powered organoid platforms to transform drug development. The company is also building an organoid biobank to support drug discovery and reduce reliance on animal models, which often fail to accurately predict human responses. The team aims to grow this biobank to 1,000 organoids, enabling more ethical and effective pharmaceutical testing.

As ProxyBio pursues FDA approval for its ProxyPATIENT offering and works toward its goal of one-week screening times, the company exemplifies how AI and advanced biological models can address one of drug development's most persistent challenges: predicting which therapies will actually work for individual patients before expensive and time-consuming human clinical trials begin.