‘Datathon’ demonstrates AI’s potential, identifies 5 repurposable drug candidates

By Melissa Fassbender contact

- Last updated on GMT

(Image: Getty/monsitj)
(Image: Getty/monsitj)

Related tags: AI, Artificial intelligence, Data

Datathon participants identify five repurposable drug candidates, applying AI to identify genes of interest and support target identification – a task that previously would have taken years, says industry expert.

The Pistoia Alliance and the information analytics business Elsevier recently held a “datathon” with the goal to identify repurposable drug candidates to treat chronic pancreatitis, a rare disease with no known cure.

Participants from various organizations, including life sciences as well as technology and academia, successfully identified five drug candidates for repurposing.

The project was sponsored by the top 20 pharmaceutical companies and conducted in partnership with Cures Within Reach and Mission: Cure, which is considering the candidates’ potential to proceed into patient trials after passing an expert review panel.

All participants had access to a life science-specific artificial intelligence (AI) environment through Elsevier’s Entellect platform, explained Dr. Jabe Wilson, consulting director, text and data analytics, Elsevier.

Elsevier in December 2018 launched​ the cloud-based data platform to de-silo, contextualize, and connect drug, target, and disease data. The datathon marked the first public trial of Entellect.

According to the company, the platform is designed to help life sciences companies overcome research and development (R&D) challenges by organizing data and presenting it in an AI-ready environment.

“As part of the datathon, participants applied AI and statistical techniques to identify genes ‘of interest’ using open-source datasets, and to support target identification,”​ Wilson told us. Participants evaluated approximately 2,000 repurposable drugs to predict possible combinations and adverse effect profiles.

Wilson said, “The speed at which this analysis could be carried out using AI is what makes all the difference, what would previously have taken years, took between 30-60 days during the datathon.”

The event was held with two main goals in mind, first, to identify drug candidates for repurposing and “to show that AI and machine learning, when applied to clinical, pharmaceutical and biochemical data, has the potential to support the decision making of subject matter experts,” ​Wilson explained.

“Each participant was able to use Entellect to go from exploratory data analysis and data preprocessing, to feature engineering, model building, validation and comparison, and finally result visualization and model deployment,”​ she added.

Participants had access to Elsevier’s Reaxys, PharmaPendium, and Pathway Studio datasets and also were able to upload their own.

“This datathon was able to show that by working together, pooling industry expertise, and using innovative technology such as AI, we can find already approved drugs with the potential to treat chronic pancreatitis,”​ said Wilson. “Not only is this much more financially feasible than developing new therapies from scratch, it has the potential to bring hope to millions of patients around the world.”

Full results of the datathon are slated to be announced at The Pistoia Alliance​ ‘Centre of Excellence for AI/ML in Life Sciences’ workshop in London on March 12th​.

Related news

Show more

Related products

show more

Stability Testing As a Quality Control Measure

Stability Testing As a Quality Control Measure

Frontage Laboratories | 19-Jun-2019 | Technical / White Paper

Stability testing is, in essence, a quality control process and is there¬fore a vital component of every phase of clinical development for both large and...

Related suppliers

Follow us

Products

View more

Webinars