Transforming life sciences and health care with graph technology

Life sciences and health care involve increasingly vast amounts of data that contain valuable insights to improve health and treatment outcomes. But storing, mining and analysing this information is increasingly beyond the scope of traditional databases and analytics tools.

This is where graph technology can play a critical role. It can help life sciences users — pharmaceutical companies, chemical manufacturers, biotech startups and healthcare providers — integrate and analyse large, complex datasets, leading to better drug discovery, disease diagnosis and personalised treatments.

In traditional data systems, data is typically organised into tables with rows and columns. This works for storing and querying structured data but struggles with complex datasets. Graph technology is a specialised data system representing data points as nodes and links to show their relationships. They’re ideal for querying huge, highly interconnected data, making it possible to traverse relationships quickly and answer complicated queries.

This presents a range of applications for the healthcare sector that can help solve complicated life sciences problems at every scale in the following areas:

1. Drug discovery

In traditional drug discovery, researchers typically screen large libraries of compounds against a single target protein or pathway. This often fails to consider the complex interplay between various proteins and pathways within the body. With graph data science, an analytics and machine learning (ML) solution, the complex interactions between biological molecules can be modelled and analysed, making it easier to identify potential drug targets and predict the effects of different drugs.

GSK, formerly GlaxoSmithKline, is leveraging knowledge graphs to improve clinical reporting workflows, answer critical scientific questions and manage drug discovery projects. A knowledge graph connects and contextualises various data structures and formats to unearth relationships and make more informed predictions and decisions.

GSK spends tremendous time processing data through its life cycle, from initial collection to study and reports and, ultimately, submission. Much of this data has traditionally been held in isolated silos, in complex programmatical scripts, making it very difficult for non-programmers to understand and for researchers to connect and analyse. Instead, GSK is moving away from isolated data domain silos to a single, contextualised Clinical Knowledge Graph that anyone can easily follow.

2. Precision medicine

Graph data science can enable much more personalised and targeted treatment plans based on a patient’s unique genetic and biological makeup. A graph can be constructed of an individual’s genomic data and analyse data from other sources such as electronic health records and wearable health tracking devices. This can help identify risk factors and deliver highly personalised treatment plans.

In GSK’s case, its patient-centric data graph will integrate all these different sources, providing rich contextual knowledge into an individual patient’s situation. This will lead to accelerated decision-making — critical for optimal patient care — and enables greater control over data privacy and patient consent.

3. Disease diagnosis

Graph technology offers a more holistic approach to disease diagnosis, making it easier to analyse biological data and uncover patterns and connections between biological markers and symptoms. For example, by building a graph of gene expression data from patients with a specific disease, researchers can identify gene expression patterns associated with the disease and key genes and pathways that may be involved.

DZD, the German Center for Diabetes Research, uses graph technology to bridge different datasets and find treatment measures for diabetes across multiple disciplines. Scientists can examine the disease from many different angles to better understand diabetes’ causes, asking questions in natural language such as: “Can a specific change be attributed to a healthier diet, a drug or a hitherto unknown factor?”

4. Healthcare analytics

Health data comes from various disparate sources, such as electronic medical records, claims and patient-generated data. These can be connected in a graph along with other data sources, such as social determinants of health or environmental factors. This may reveal potential links between these factors and possible interventions to improve health outcomes by addressing the underlying factors.

Novartis has used graph technology to combine its data with medical information from the National Institutes of Health’s PubMed. It contains 25 million abstracts from around 5600 scientific journals, connecting linked genes, diseases and compounds in a triangular pattern. Researchers can then make queries such as: “I want to find compounds similar to this compound that have annotations about this disease”.

Graph technology is a powerful tool for life sciences analytics that allows researchers to model and analyse complex interactions between different variables and outcomes in healthcare data. By uncovering patterns and connections that may not be apparent through traditional approaches, graph technology can result in more targeted and effective healthcare interventions, improving patient outcomes and reducing costs.

Top image credit: iStock.com/NicoElNino