Amii helps Edmonton-based startup validate projects and increase in-house technical capabilitiesOverview

DrugBank is one of Canada’s fastest growing health tech companies. In the past year, they have grown exponentially and focused on building products powered by its curated and structured pharmaceutical knowledge database. DrugBank provides data solutions for healthcare companies working in precision medicine, drug discovery, drug repurposing and clinical software applications. Its publicly available resource, DrugBank Online, has been cited in over 12,000 academic publications and is used by millions of researchers and health professionals globally.

In addition to serving health care professionals, DrugBank offers products such as structured clinical and scientific data to some of the most prestigious organizations and name brand pharma companies on the globe, such as Healx, Aria Pharmaceuticals, and Molecular Health. Since its inception in 2006, DrugBank has grown a lot, starting as a research project at the University of Alberta, entering commercialization in 2015, and is now the world’s largest pharmaceutical knowledge database.

“The future of medicine and drug discovery is data-driven,” says Chief Technology Officer Craig Knox. “DrugBank is in a unique position; the core technology consists of this highly-structured and rich database, and what’s unique about it is that it offers both clinical and scientific information. Clinical information is things like side effects and drug interactions. Scientific information is how the drugs actually work in your body.”

DrugBank began its engagement with Amii in 2018 when they received their initial AI feasibility assessment. “We started using Natural Language Processing (NLP) almost immediately when we built it out,” says Chief Executive Officer Michael Wilson. “It’s always been a part of DrugBank and we just keep putting more and more focus on it.”

With Amii’s help, DrugBank hired their first NLP specialist in 2019, who pushed their machine learning (ML) project forward, providing a big win for their internal data and growing NLP expertise within the company.

Approach

Prior to their engagement with Amii, DrugBank was at the Exploring stage of Amii’s AI Adoption Spectrum. At the end of their engagement, DrugBank entered the Implementing stage.“We never wanted to position ourselves as strictly an AI company when we started out as a commercial project,” says Knox, “but it’s been an ongoing development and it’s very important in how we use that technology in the future.”

Through their engagement with Amii, DrugBank has:

1. Validated projects. In order to de-risk and future proof their investments in a rapidly growing field, DrugBank worked with Amii to help validate their existing and future projects when building out their tools.

2. Hired an NLP Specialist. Amii and DrugBank worked together to write the job posting, screen applicants and perform technical interviews with candidates. Working together allowed DrugBank to develop their internal knowledge and capabilities regarding hiring.

3. Integrated with the AI community. Having the opportunity to connect with people who have similar interests in AI and ML -- from both the academic and commercial world -- DrugBank has participated in a variety of talks and meetups to help raise awareness, support and grow the AI community, and their business within it.

Results

“Our core technology, the knowledge base, which includes AI for quality control of data processing is integrated in so many aspects of the company,” says Wilson. “It’s not this nice little package that someone can provide us a deliverable and we plug it in. The level of value we can get from building it into the core of our company is much higher.”

A drug database in a clinical and scientific world needs two essential criteria: accuracy and coverage. In terms of accuracy of data, the goal for DrugBank is to get as close to 100% accuracy as possible, but the coverage is a difficult task since there are many academic papers being published every day.

“It’s challenging to keep up manually, unless you have tens of thousands of curators trying to stay up to date with that information. So, we use AI internally to start to extract a lot of this information to improve our coverage,” says Knox. “By using NLP tools, the models can extract that information into the hands of the curation team so they can go through the last mile. Then we can say: we can feel confident the accuracy and coverage for this particular data set is close to 100% as we can possibly get.”

Conclusion

Through their engagement with Amii, DrugBank has amplified the company's commitment to AI and ML, allowing them to grow their technical team and increase their capacity to communicate about their ML concepts and projects effectively.

Since 2018, DrugBank has quadrupled the team, and closed a significant funding round. In the coming years, Michael Wilson and Craig Knox plan to grow the technical capacity within DrugBank, while investing in AI/ML talent for their company.

“I think Amii has helped build the AI reputation of Alberta and Edmonton as a whole, which has an effect in improving the way people view our company,” says Knox. “We’ve had examples in the past where people are familiar with Amii or the affiliated researchers, and when it’s known that we have direct connections there, it’s helped us gain validity on the commercial side of our business.”

To learn more about DrugBank’s product offerings, please visit drugbank.com.