Skip to main content

Cloud AI enabled analysis for cancer treatment development

The Deloitte AI Institute, Deloitte’s global centre for AI research and applied AI innovation, recently worked with Arjuna Therapeutics to build an AI-driven research engine. Arjuna Therapeutics is an emerging biotechnology company developing an entirely novel class of small molecules, Therapeutic Molecular Clusters (TMCs), for the treatment of various cancers. Arjuna’s vision is to create the most effective treatment for cancer patients. As such it is developing a pipeline of molecules by combining novel methods of action and biomarker-driven patient treatment with Machine Learning embedded at the heart of its research.

Arjuna’s pre-clinical research involves combining data from internal and external sources, applying statistical techniques, and communicating results to expert and non-expert stakeholders. Data from the research includes relationships and measurements of thousands of genes, proteins and cells. Traditional statistical techniques can often struggle to separate signal and noise when compared with AI techniques for pattern recognition on high-dimensional data. Arjuna wanted to develop an AI-driven research engine and needed a platform to make this approach more efficient, robust, and scalable with respect to additional data sources, analysis and visualisation methods, and a growing pipeline of molecules.

Cloud and Machine Learning specialists from the Deloitte AI Institute, along with bioinformatics experts from Deloitte’s Life Sciences Practice, worked with Arjuna to understand and refine the existing approach using AI. The team implemented data, AI and solutions engineering best practices, including automated data validation, feature engineering pipelines and a centralised database. They also integrated open source code from both the bioinformatics and AI ecosystems. This allowed for the exploration of AI techniques such as dimensionality reduction as well as supervised learning, bringing in an alternative data science perspective to enhance insights coming from the data. Preliminary results, in terms of statistical associations between molecular treatment effectiveness and biological signals at the gene, protein and cell levels, are promising and suggest potential focus areas for analysis of the pipeline of molecules.

The team are excited for what comes next. With the current platform, Arjuna will be able to more efficiently and effectively identify and analyse candidate treatment molecules, and communicate these findings. The platform will serve as a foundation upon which incremental data sources, analysis and visualisation techniques can be added. The Cloud design means that it can scale as required. The platform also has potential value for the Life Sciences industry more broadly. The open source integrations include features commonly used in R&D, for example dose-response curve fitting and gene-set enrichment analysis, which are now packaged in a robust AI-driven Cloud platform which is easy to deploy and adapt. Against an industry backdrop of recent AI world-firsts including human trials of drugs discovered by AI1 and state-of-the-art protein-folding analysis2, the team sees increasing interest and opportunity for applying AI within Life Sciences innovation. These organisations are increasingly looking to scale the technology and infrastructure needed to support their expert talent, and our collaboration illustrates what can be achieved as a result.