Great Ormond Street Hospital has unveiled a new platform designed to generate clinical intelligence from routine health data, developed in-house by its Data Research, Innovation and Virtual Environments (DRIVE) unit.
The trust highlights that whilst EPRs have enabled hospitals to capture a “huge amount” of information about patients and their conditions, “there is no consistent way to combine and analyse this data for broader clinical or operational insights”.
In response, the Paediatric Informatics Consultation Using Real-world Evidence (PICTURE) platform was launched, supporting the aggregation and analysis of EPR data. According to a study published by the trust’s team, aims were to allow users to define arbitrary patient cohorts, provision EPR data for those cohorts, offer “extensive analytics functionality”, present outputs “conveniently”, and “maintain applicability across various healthcare organisations and settings”.
Developed using open-source languages including R and Python, the platform takes de-identified data from patient cohorts and transforms it into “a bespoke in-house common data model” developed as part of a pre-existing ETL library, that outlines clinical events like diagnoses and procedures in table format.
A library of analytics functions such as frequency analysis and measures-of-association analysis, then offer clinical intelligence in response to input queries. “Crucially, if analytic functionality is required that does not exist in the library, it can be created using an open-source programming language and integrated with PICTURE,” the study team shares. The platform is also designed to allow for the incorporation of AI and machine learning services operating via cloud or third-party intelligence visualisation tools.
Suggested use cases cover keeping patients informed, enabling research, supporting hospital operations, and improving clinical practice. Looking ahead, the GOSH DRIVE team suggests the potential for use in “multi-site, federated analytics” whereby standardised analysis packages could be “distributed and executed locally at collaborating institutions, enabling large-scale research without centralising sensitive patient data”.
Digital at GOSH
GOSH released its first ever AI strategy, highlighting the “unprecedented opportunity” AI presents for revolutionising healthcare delivery and enhancing clinical outcomes, driving efficiencies, reducing clinician burden, and improving patient care. Years one to three will focus on scaling and optimising AI use, including in areas such as diagnostics, operations, and financial management. An AI monitoring system will be developed to track performance and ensure safety. AI research collaborations will be sought with NHS, industry, and academia, and the trust will work toward HIMSS AMAM Level 7.
The trust shared digital achievements around ambient voice rollout, EPR training, and AI, highlighting plans and commitments in line with the 10-Year Plan. A recent internal audit benchmarking digital and AI capabilities across 22 different trusts “concluded GOSH was significantly ahead”, the trust notes. The launch of Bedside MyChart, an inpatient portal, achieved rapid adoption, with 38 percent of inpatients engaging in the first month, saving more than 24 hours of nursing time. October 2025 also saw the implementation of the second phase of the London Care Record, reportedly enabling visibility of patients across other London trusts.
GOSH also recently shared plans to expand its pilot of ambient AI to 5,000 patients from a range of different healthcare settings across London, following testing in 100 outpatient appointments. After “detailed and thorough safety checks to ensure that the technology is secure” and early phase testing which saw the AI assistant trailed by clinicians, medical actors, and 100 real patients, GOSH is hoping that TORTUS can help increase face-to-face time during appointments, supporting staff and delivering improved outcomes for patients.
