Dexter AI: from research tool to ambitious spin-out

These hold huge untapped potential for health researchers, containing evidence about how treatments or interventions perform in everyday clinical practice rather than under the highly controlled conditions of a clinical trial. But the data in EHRs is often imperfect, or incomplete, and extracting it in a usable form is labour-intensive work.

Professors Krishnarajah Nirantharakumar and Joht Singh Chandan, and computer scientist and researcher Dr Krishna Gokhale from Birmingham’s Department of Applied Health Sciences set out to automate this process. This evolved into a platform that can perform in seconds tasks that would previously have taken weeks to carry out.

The Dexter AI software extracts analysable datasets from EHRs across primary and secondary care, claims data, private prescriber records, and other sources, turning Real World Data into Real World Evidence (RWE) that can be used to address questions posed in epidemiology, pharmacovigilance, and health economics studies.

DexterRWE has gone on to support over 150 peer-reviewed studies, identifying associations, trends and risk factors that link common conditions.

The bigger prize, however, was working directly with the NHS. Dexter AI’s second product, DexterCare, allows healthcare professionals to automate healthcare audits, and this enabled the team to complete projects with the NHS as well as multinational pharmaceutical companies. 

Its first use case was at the Shrewsbury and Telford Hospitals NHS Trust, where DexterCare drastically accelerated and streamlined clinical audits to improve maternity outcomes.

Where previously a midwife had spent several days manually interrogating a complex EHR system to audit 10s of patient records, DexterCare automated the process and applied it to 1000s of patient records in under 10 seconds.

Before it spun out, University of Birmingham Enterprise incubated Dexter AI through its unique Operating Division (OD) framework. During this phase, the Enterprise team provided operational support, handling the legal side of early commercial agreements with major clients, as well as shouldering the insurance and accounting responsibilities.

The OD framework allows potential spin-outs to mature, understand the market, and hone their approach. In Dexter AI’s case, it allowed the team to build partnerships with industry players while evolving the offering into an industry-ready platform tested across diverse data types.

The incubation period allowed the Dexter team to develop a web-based application that users can take to their own data, which means data can be extracted and analysed securely without the risks involved in moving data from one platform to another.

Dexter AI also completed projects for IQVIA, Cegedim Health Data and Adelphi Real World. Dexter can reduce the time and cost of RWE studies, identify suitable participants for clinical trials from EHRs, and conduct population health-based research.

Dexter AI plans to take the audit tool to acute NHS healthcare settings, such as emergency care and cardiology departments. It has ambitions beyond healthcare, and is looking to take the DexterRWE tool beyond medical health records, to analyse large datasets from social care, mental health services and crime databases.

Although the team has delivered projects using secondary care datasets and primary care data from across Europe, Dexter AI remains physically and operationally anchored in Birmingham.

The entire Department of Applied Health Sciences adopted the software and continues to use it for observational health studies. The Dexter team has also forged a deep partnership with University Hospitals Birmingham, which is using Dexter AI to rapidly extract data from maternity and acute care records.

Professor Nirantharakumar, now based at King’s College London, is Dexter AI’s Chief Scientific Officer, Dr Gokhale its Chief Technology Officer, and Professor Chandan, who became the UK’s youngest Clinical Professor in 2024, its Chief Medical Officer. The company is led by CEO Christian Billinghurst, who worked with the team while Dexter AI transitioned from an academic project to a commercial venture.