Predictive Oncology Inc said it has achieved “a key milestone” towards monetizing its unique approach to leveraging artificial intelligence (AI) in oncology drug discovery.
Earlier this year, Predictive Oncology’s Discovery 21 proof-of-concept campaign demonstrated its Patient-centric Discovery by Active Learning (PeDal) platform’s ability to make high-confidence predictions of drug responses on ovarian tumor samples.
The Minneapolis, Minnesota-based company, which is focused on applying AI to develop personalized cancer therapies, said the results from Discovery 21 are now validated and PeDAL has “accurately and reproducibly” predicted drug response results.
READ: Predictive Oncology set to market its flagship artificial intelligence drug discovery platform
“These results demonstrate that PeDAL works with real-world drug compounds and tumor samples,” Predictive Oncology CEO J Melville Engle said in a statement. “We expect that pharma will be keen to use this technology, which increases the probability of technical success and lowers both the cost and time of bringing new drugs to market. This in turn will bring us closer to our overall mission of determining optimal therapies for the treatment of cancer.”
Discovery 21 tested PeDAL’s ability to make drug response predictions, starting with an experimental space of 175 US Food and Drug Administration (FDA) approved cancer drugs and tumor samples from 130 ovarian cancers.
“Predictive Oncology has a database with drug response results for over 150,000 tumor samples in over 137 tumor types,” said Pamela Bush, senior vice president of Strategic Sales and Business Development, at Predictive Oncology. “When we combine our database with PeDAL’s proprietary AI, we have an impressive tool for predicting whether a tumor will respond to a particular drug.”
In Discovery 21, PeDAL’s AI predicted whether a specific tumor sample would respond to a drug with 91.8% accuracy, including predictions on drug/tumor sample combinations that had not been tested before. The Discovery 21 team has also developed a customer-facing portal to allow partners to evaluate and analyze project data and outcomes in real-time. The visual output of the portal will provide a platform for active collaboration and decision-making.
The company’s PeDAL platform is powered by machine learning that drives experimental testing to evaluate models of hundreds of diverse tumor samples against hundreds of drug compounds for early-stage drug discovery. PeDAL’s AI iteratively improves the predictive models until it reaches a specific pre-determined goal. The customizable platform can leverage incomplete data and efficiently pick the best set of experiments so it can learn from the new experimental data.
Predictive Oncology explained that by selecting key experiments, PeDAL’s AI can make high-confidence predictions without testing the entire experimental space – meaning every drug with every tumor sample – thus saving valuable time and costs for the pharma industry. It inevitably helps companies select better drug compounds to move into clinical development. The AI component of PeDAL comes from an exclusive world-wide license from Carnegie Mellon University.
The completion of Discovery 21, and its subsequent validation, creates a path for Predictive Oncology to monetize its AI-driven approach, which brings efficiency to the drug discovery process. Predictive Oncology is set to be a first-mover in the AI-powered drug discovery market that the company estimates will grow to $20 billion in the next three years.
Contact the author Uttara Choudhury at uttara@proactiveinvestors.com
Follow her on Twitter: @UttaraProactive