The Clinical Record Interactive Search (CRIS) system is a computer system, developed in 2008, that allows researchers at the NIHR Maudsley Biomedical Research Centre (BRC) to carry out research using information from the South London and Maudsley (SLaM) NHS Foundation Trust clinical records. The Trust serves a local population of nearly two million people across four London boroughs. It has more than 230 services including inpatient wards, outpatient and community services. It provides inpatient care for over 5,000 people each year and treats more than 45,000 patients in the community in Lambeth, Southwark, Lewisham and Croydon. CRIS provides authorised researchers with regulated access to anonymised information from the clinical record. CRIS is safe and secure and researchers who use it cannot access patients’ personal details. Applications to access CRIS and the analyses carried out using CRIS are closely reviewed, monitored and audited by a CRIS Oversight Committee, which carries representation from the Maudsley Caldicott Guardian and is chaired by a service user. The CRIS Oversight Committee is responsible for ensuring all research applications comply with ethical and legal guidelines. The system consists of a series of data-processing pipelines, which extract anonymised data from structured fields as well as unstructured free text from case notes and correspondence of the clinical record. It contains over 350,000 de-identified patient records, which in turn represent over 30 million free text documents growing at a rate of about 150,000 documents per month.
CRIS helps us to look at real life situations on a large scale over time. This means it's easier to see patterns and trends, like what treatments might for some people but not for others. Data in CRIS are used in:
SLaM audit and service evaluations
Observational research
Recruitment to research studies
The Natural Language Processing (NLP) Service One of the techniques that have been implemented in CRIS is NLP. Essentially, what we have achieved is the development of a language model for the automatic extraction of information from the free text in the electronic health record. We are currently using almost 100 such automatic algorithms that enable us to automatically search the free text for a variety of symptoms, interventions, predisposing factors and outcomes.
