Atlas Discovery launches AI models to predict responses in clinical trials (runtimewire.com)

Atlas Discovery has emerged from stealth mode, launching innovative foundation models aimed at predicting individual patient responses to drugs in clinical trials. Co-founders Shaamil Karim, Sanjukta, and Christian aim to bridge the costly gap between preclinical findings and effective human treatments. By utilizing a combination of abundant preclinical data and scarce clinical insights, their models have shown promise in applications such as drug repurposing, identifying suitable trial patients, and evaluating in-licensing opportunities. In retrospective analyses of cancer and autoimmune trials, Atlas's models distinguished between responders and non-responders with impressive accuracy, notably predicting treatment responses in a phase 3 trial of ustekinumab. This development is significant for the AI/ML community as it addresses a major challenge in drug discovery: transferring insights from preclinical models to human responses, which has historically hindered progress. The company's approach not only highlights a novel methodology in machine learning for drug response prediction but also offers potential cost savings in trial design, as illustrated by their case study suggesting that participant numbers could be reduced without sacrificing statistical power. With plans to forge partnerships with healthcare institutions and biopharma teams, Atlas Discovery stands at the forefront of applying AI-driven solutions to revolutionize clinical trial efficacy and drug development.