++Algorae Pharmaceuticals (ASX: 1AI)++ has launched a major upgrade to its AlgoraeOS AI-driven platform for the discovery of drug combinations.
Researchers at the University of New South Wales Biomedical AI Laboratory and the UNSW AI Institute helped the company develop the upgrade, with additional support from CSIRO Data61.
Changes to the cutting-edge platform address key limitations of the original version and set a higher standard for the identification of effective combinations to treat cancer and other complex diseases.
Outperforming Industry Models
AlgoraeOS v2 outperformed representative state-of-the-art models – including Google DeepMind’s TxGemma-27B-Predict and Tx-LLM (M) – in published benchmarks, while also demonstrating stronger calibration across biologically-diverse, clinically-relevant synergy regions.
The platform has been trained at scale on over 5.5 million unique inhibition records from harmonised high-throughput combination screens and engineered to model the full dose-response surface across numerous independent drug analysis models.
The results provide Algorae with a more reliable, decision-grade tool for prioritising combinations, selecting doses, and designing preclinical studies.
The company expects to release the first in silico fixed-dose combination predictions before year end.
Predictive Insights
Executive chairman David Hainsworth said AlgoraeOS v2 would significantly enhance the company’s pre-clinical development pipeline by providing predictive insights to guide drug candidate selection and design.
“The launch of our platform upgrade is a significant milestone for Algorae which positions artificial intelligence as the cornerstone of our research and development programs,” he said.
“Across published benchmarks, this new version has delivered clear performance gains over representative state-of-the-art models, which is a major achievement in our dual-track strategy of AI-driven innovation and pharmaceutical commercialisation.”
New Generation of AI Models
“AlgoraeOS represents a new generation of AI models that predict outcomes and understand their own limits,” UNSW associate professor Fatemeh Vafaee said.
“By combining biological knowledge with uncertainty-aware deep learning, we can now model drug interactions with greater reliability and generalisability than ever before.”
UNSW post-doctoral research associate Dr Muhammad Javad Heydari said AlgoraeOS would help advance drug discoveries.
“This upgrade is designed to analyse combination therapies the way a modern lab approaches them — by integrating mechanisms, context and dose in an advanced AI model,” he said.
“It turns large-scale biological data into clear guidance on which drug pairs to advance, at what doses and with quantified risk for each decision.”
