needs to be progressively refined so that those variables which best correlate to the targeted vulnerability are used. This is consistent with the data protection principle of data minimization and should reduce false positives., i.e., inappropriate identification of an individual or household that are not vulnerable.

In developing, and improving, a Risk Stratification Policy, a

@Lead Organisation⁠

will need to draw upon a number of sources of knowledge, information and data, for example

Knowledge derived from studies that identify relationships between specific indicators and a vulnerability – these studies published by academics, charity/third sector organizations or public authorities may provide evidence of how good a

@Vulnerability Attribute⁠

is when trying to predict or identify a specific vulnerability.

Data collected during an earlier SAVVI process can be depersonalised to find which combination of

@Vulnerability Attributes⁠

were most successful.

People who were identified during a

Find⁠

phase, or asked for assistance during the

Support⁠

phase, can be later contacted to better understand their profile to further refine the choice of

@Vulnerability Attributes⁠

Information about the claimed and actual data quality of data sources, e.g. their scope, coverage, means of collection and verification. This is helpful in assessing how effective use of a particular dataset may be and its limitations.

There are a range of advanced analytical techniques, for example clustering and evolutionary computing, which can test millions of algorithmic combinations of input variables and score each and rank them to see how well they fit with the real world. Where these techniques are considered, specialist advice should be sought and ethical considerations taken into account.

Developing algorithms for all but the simplest relationships potentially requires research and data analytic skills and may best be assigned to organisations with secure trusted analytic environments where data from multiple sources can be investigated and modelled.

Risk stratification is a form of profiling and subject to specific legal constraints in the Data Protection Act 2018 and UK GDPR. The formulation of a Risk Stratification Policy should be underpinned by a formal evidence base, including but not exclusively literature reviews on the topic and the types of study referred to above that have contributed to an understanding of the vulnerability and its potential causes and indicators.

Load content from www.digitalservicedesigner.com?