We are determined to advance our fundamental understanding of diseases such as cancer, respiratory disease and heart, kidney and metabolic diseases. Because by learning what causes or drives disease, we hope to find new ways to treat, prevent or even cure them.

Using graphs to turn knowledge into insights
Knowledge graphs are networks of contextualised scientific data facts and the relationship between them. Our knowledge graphs integrate genomic, disease, drug, clinical and safety information, helping to overcome confirmation bias and to turn data into insights. Machine learning and AI applications such as graph neural networks can then mine this data to uncover previously unknown patterns and make novel target predictions. In 2021, we selected the first two AI-generated drug targets into our portfolio, from our collaboration with BenevolentAI. We share parts of our internal knowledge graph work on GitHub.  

Unlocking secrets in our genes and beyond
Our Centre for Genomics Research is working towards the analysis of up to two million genomes by 2026. We use best-practice cloud environments to process and apply advanced data and AI tools to interpret the vast genomics data faster and more robustly than previously possible.

Beyond the genome lie the dynamic realms of the transcriptome, proteome and metabolome – largely untapped repositories of rich information that if connected could tell us more about what is driving disease. Multi-omics is the integration of these datasets which, with the help of machine learning and AI, can help us predict what a drug molecule does in a cell with far greater certainty.

Every week, our pathologists analyse hundreds of tissue samples from our research studies. They check them for disease and for biomarkers that may indicate patients most likely to respond to our medicines. It is very time consuming which is why we are training AI systems to assist pathologists in analysing samples accurately and more effortlessly. This has the potential to cut analysis time by over 30%.

For one of our AI systems, we implemented an approach inspired by how some self-driving cars understand their environment. We trained the AI system to score tumour cells and immune cells for a biomarker, called PD-L1, which has potential to help inform immunotherapy-based treatment decisions for bladder cancer.

Cancer is not the only disease where imaging and AI are transforming research. Recently one of our biopharmaceuticals research teams undertook an ambitious project to train deep neural networks to predict disease risk and related biomarkers from retinal fundus images.  

Randomised Clinical Trials (RCTs) are currently the method of choice for pharma when it comes to assessing potential new medicines. However, published data shows they have become more expensive and complex over time.

Advances in data science can help us re-think clinical trials, enhancing current practice and finding new ways to discover and develop potential new medicines.

For example, the rapid adoption of high-quality Electronic Health Records (EHRs) represents a vast, rich, and highly relevant data source that has a huge potential to improve clinical trial implementation.

Federated EHR technology is unlocking new opportunities to enhance clinical research and transform the way we do clinical trials. The technology has the potential to refine or replace many clinical trial processes including patient identification, selection, trial conduct, and capture of data.

We are also employing AI and machine learning tools to glean more value from clinical trial data. Historically, we have been proficient in using data from trials to analyse, interpret and report on the safety and efficacy of the trial drug. But we want to maximise the value of the data we have already collected.

Machine learning and AI are also being applied for event adjudication in clinical trials to enable us to optimise the process at different stages with the intent of reducing the time overall. 

Data re-use can help us better design our drug development strategies and programmes. This can help us design smarter trials, strengthen our scientific discoveries, and ultimately, in the future, has the potential to help our patients receive the best treatments.

Today we are generating and have access to more data than ever before. Data and analytics have the potential to transform our business, but the true value of scientific data can only be realised if it is “FAIR” - Findable, Accessible, Interoperable and Reusable.

AstraZeneca’s R&D and IT groups are working closely together to create an industry-leading enterprise data and AI architecture. This will help us answer key business questions and enhance our ability to harness new tools and technologies, such as AI and machine learning, both now and in the future.

We are also mobilising a team of data scientists, bioinformaticians, data engineers and machine learning experts from across the company to ensure we are collecting, organising and using the right data in the best way.

Rapid developments in AI technology have brought us in to uncharted territory, and companies and regulators must work together to meet the new challenges posed. Our principles will empower us and our partners to navigate this new environment safely and effectively. By encouraging innovation and evolution while maintaining our values, they provide a long-term ethical foundation to uphold our AI governance.

During 2020, we engaged a diverse range of experts both inside and outside AstraZeneca to develop principles for ethical data and AI, aligned with our Code of ethics and values. These values work for patients and employees and enable AstraZeneca to make a positive contribution to society.