Cancer from the inside out: using virtual reality to create certainty in clinical decisions using complex genomics

Background: The human genome is complex. It is within this complex data that information will be found that identify patients as individuals, the very basis of ‘precision medicine’. Functionally, this complex genome underpins the biological mechanisms of an individual patient's cancer.

Aim: To build a framework for personalised treatment of disease, where the complexity of the genome is negotiated in meaningful and actionable ways.

Methods: Machine learning analysis of complex genomic data embed the high dimensional features into low dimensional (3D) space to visualise the interrelationships between patients (eg PCA, tSNE, UMAP). Here we present VROOM (Virtual Reality to Observe Oncology Models), a novel VR prototype that allows the complete immersion of the user within the data 3D model for a cohort of patients which allows for comparative data analysis, visualisation and clinical interpretation of individual patients.

Results: VROOM is built on strong design principles and practical needs that allow analysts to move into the 3D environment to explore within the cohort for individual patients of interest. Tested on models involving gene expression data derived from 400 acute myeloid leukaemia patients, users can select individuals to compare within the 3D virtual genomic world. The system allows for the extraction of patient specific gene expression values which are then transferred to a virtual 2D working board that displays patient-specific information as well as different graphical representations, teasing out all user-defined data comparisons for the individuals in question. Stakeholder usability studies highlight ease of use and the unique merits of the application.