Opyl Limited announced that its clinical trial prediction and protocol design software solution, TrialKey, has successfully achieved a key development milestone, reducing bias to improve predictive accuracy in the model, by accurately tracing a random subset of clinical trials reported to global registries through to their outcomes, with 92% accuracy. This is significant as only around 13% of the circa 431,000 clinical trials1 registered on the US
government website, ClinicalTrials.gov, have posted their results to the site and no trials are linked over time as they progress between clinical phases. The huge information gap is a major problem for the medical research sector as no one benefits from past learnings and often repeat protocol (trial plan) design mistakes made in previous trials. This information asymmetry contributes to high trial failure rates and the unnecessary waste of time and money on research destined to fail. The scalable TrialKey Software-as-a-Service (SaaS) solution overcomes this issue by using machine learning/AI and natural language processing (NLP) to link trials across multiple global databases and sources, reducing failure bias and thereby improving predictive and protocol design accuracy and value of the software. The completion of the first major development milestone was funded under the federal government's Innovation Connections grant and was done in partnership with RMIT University's School of Computational Sciences. The linking of clinical trial outcomes across phases leading into an accurate software trial design solution will prove to be an invaluable tool for researchers around the world from big pharma to emerging biotech and medtech. With the rising costs of medical research and clinical trials that translates into the higher cost of new drugs and devices, TrialKey aims to reduce failure rates and therefore waste of valuable financial and clinical resources which it is hoped, will deliver more affordable healthcare. Using a sub-set of 17,661 phase 1, 2 and 3 trials sourced from a public registry, the Opyl team, working with Dr Antonio Jimeno Yepes and Prof Karin Vespoor from RMIT University, was able to estimate linkages to future phases for the same trial 45% of the time. For example, it shows that of the 6,847 phase 2 trials conducted, Opyl was able to find 3,800 phase 3 trials that link to a phase 2 study (55%).