Akrivia SMARTbiomed workshop

A 2-day workshop to dive into the world of Akrivia Health’s database of psychiatric electronic health records (EHR), with a special focus on mental health data access, processing and analysis. This workshop serves as a gateway to understanding the Akrivia Health cohort and to help you learn how to best use Akrivia Health’s data in your research. This workshop is ideal for anyone curious about conducting research into mental health and dementia using UK electronic mental health records. 

The cost of the workshop is £350 for the two days (9am-5pm) which includes lunch and coffee breaks. To express interest in the workshop and to receive an invitation to the registration (and payment) link please complete this questionnaire. 

NB. Workshop attendees must be affiliated to a university that has a Data Use Agreement with Akrivia, currently only the University of Oxford.

The workshop is strictly limited to 20 participants. Places will be allocated on a first come first served basis, and last sign up is Friday 25th April.

If you are interested in attending a workshop in future but cannot make the dates, then please complete the questionnaire to help us with future planning.

The workshop is co-sponsored by the Pioneer Centre for SMARTbiomed (statistical and computational methods that advance research to transform biomedicine), a collaboration between the University of Oxford and Denmark. Please email smartbiomed_info@bdi.ox.ac.uk if you have any queries about the workshop.

Day 2: Join us to use the secondary mental health EHR dataset to answer important research questions. Direct hands on use of the data we expert advice on how to approach your research question.

Goal: Participants will come away with general understanding of the Akrivia Health data and have direct first-hand experience if Akrivia Health data are suited for their research questions. 

ABOUT THE AKRIVIA HEALTH DATASET

Akrivia Health is one of the world’s largest and most in-depth datasets, comprised of anonymised longitudinal psychiatric data derived from the electronic health records (EHRs) of over 6 million patients in secondary mental health and dementia care across England and Wales. The data is regularly refreshed and harmonised across 20 NHS Healthcare Organisations providers. Akrivia Health was initiated in 2019, with patient records dating back to before 1990, including participants with psychiatric health record entries, likely over representing individuals with severe mental illness compared to the general population. The dataset contains structured data, such as patient demographics and service pathways, while research-relevant information is also extracted from free-text clinical notes using natural language processing (NLP).

Profile paper DOI
https://doi.org/10.1136/bmjopen-2024-088166

Cohort

6,331,592 patients as of February 2025

Data source:
•               Mental health and dementia data
•               Dataset is from a single specialty and tier of care, secondary mental health care
•               20 NHS Healthcare Organisations from England and Wales

Representative sample
•               Psychiatric population of England and Wales
•               ~40% of England and Wales’ in-care population for severe mental illness and dementia

Longitudinal?
Yes (first medical records since 1990)

Ongoing?
Yes

Methods
•               Quantitative secondary data
•               Anonymised Psychiatric Electronic Health Records (EHR)

Prescription records
•               None structured at source
•               Medication information is extracted from the free text of progress notes using NLP.

Linked primary care data
•               Established data linkage with a separate primary care dataset of GP records in the UK.

Linked bio-sample data
•               Established data linkage with a whole genome sequencing, biomarkers and IPSCs for a subset of patients in key disease areas including major depressive disorder, bipolar disorder, schizophrenia and dementias.

Neuroimaging data collection
•               None

Data collected available for research

Current dataset consists of structured at source variables and NLP-derived variables.

Structured at source data:
•               demographic information
•               diagnosis
•               referrals
•               inpatient stays

NLP data:
•               medication
•               diagnosis
•               health scores
•               signs and symptoms
•               psychotherapy
•               substance use
•               ECT