United Kingdom, Thames Valley

Thames Valley Police

DARAT (Domestic Abuse Risk Assessment Tool, machine learning), supplied by Solvarithm Ltd (contractor)

In developmentOfficial publicationPAI-68632146

Body
Thames Valley Police
Country
United Kingdom
Area
Thames Valley
Technology
DARAT (Domestic Abuse Risk Assessment Tool, machine learning)
Supplier
Solvarithm Ltd (contractor)
Status
In development
Being researched, built, tested or marketed for police use. No police body has confirmed using it.
Stage
On the horizon
Since
Not known
Legal basis
Not stated in the source
Oversight
A Data Ethics Committee of public members and academics has reviewed the model three times, according to the algorithmic transparency record. A data protection impact assessment was ongoing, with an algorithmic impact assessment and an equality impact assessment planned before implementation.
Evidence
Official publication
Police force, government department or public body publication
Source
GOV.UK algorithmic transparency record, 14 October 2022
Last checked
Record code
PAI-68632146

Notes

The algorithmic transparency record for the Hampshire and Thames Valley Police Domestic Abuse Risk Assessment Tool (DARAT), first published on 14 October 2022, describes a machine learning classification tool, currently a random forest classifier, that forecasts whether harm will occur in the year after a domestic abuse incident and sorts risk into standard, medium or high. It is designed to supplement the existing DASH assessment, with a supervisory officer seeing the score and its basis. The record says DARAT "is not currently used for decision making" and that refinement, bias checking and testing remain before trial and implementation. It reports an initial model that improves overall classification accuracy by 20 percentage points over DASH and is nearly three times as sensitive to high-harm outcomes; these are the developers' figures. The responsible team is the Thames Valley Violence Reduction Unit data and targeting team, and the developer is named as a contractor, Solvarithm Ltd. The two forces have separate tools built on each force's records. No later source read shows the tool entering use.

Other systems used by Thames Valley Police

Other systems used by Thames Valley Police
DeploymentCountrySupplierSinceStatusChecked
Thames Valley PoliceLive Facial Recognition United KingdomThames Valley, first deployed in Oxford city centre — 2025 Operational
Thames Valley PoliceAI Emergency and Non-Emergency Call Triage United KingdomThames Valley Salesforce (Agentforce 360 platform), according to Salesforce 2025 Operational

DARAT (Domestic Abuse Risk Assessment Tool, machine learning) elsewhere in United Kingdom

See them in the tracker
DARAT (Domestic Abuse Risk Assessment Tool, machine learning) elsewhere in United Kingdom
DeploymentCountrySupplierSinceStatusChecked
Avon and Somerset PolicePredictive Analytics Programme United KingdomAvon and Somerset Built in-house 2017 Operational
Bristol City CouncilThink Family Database and child exploitation risk models United KingdomBristol Not specified 2016 Discontinued
Ministry of JusticeIndividual Risk Scoring United KingdomEngland and Wales (research phase) Developed in-house (Ministry of Justice data science team) 2023 Proposed
Essex PoliceKnife Crime and Violence Model (Fearless Futures) United KingdomEssex — — Trial
Lincolnshire PoliceCESIUM child exploitation risk identification United KingdomLincolnshire Trilateral Research (CESIUM) 2023 Operational
Nottinghamshire PoliceCESIUM child exploitation risk identification United KingdomNottinghamshire Trilateral Research 2024 Trial
West Midlands PoliceStalking and harassment harm escalation model United KingdomWest Midlands — — Operational
West Yorkshire PoliceDomestic Abuse Assessment Tool (machine learning risk prediction) United KingdomWest Yorkshire — — Proposed
Metropolitan PoliceStalking Threat Assessment Centre machine learning triage United KingdomLondon — — Operational
Metropolitan PoliceGangs Violence Matrix (individual risk and harm scoring database) United KingdomLondon — 2012 Discontinued