Population Health Data science

Within our role in One London, we are working with London’s ICBs to deepening level of insights from data, while aligning to local requirements.

We are Sharing and “porting” a central library of code, tooling, and predictive analytics models, for local adaptation and use across organisational boundaries: ‘Infrastructure-as-Code’.

Learning Health Systems:

As part of this work - we developed detailed segmentation model, with more than 60 definitions of clinical conditions, complications, and measurements, in primary and secondary care. Used to construct a descriptive, geo-spatial profile of comorbidity and resource utilisation across neighbourhood team, GPs and hospitals.

Multi-Dimensional Comorbidity Profiling

• We deploy live statistical and machine learning approaches to help population health teams target populations who are most at need.
• We use modified Cambridge Comorbidity Scores to model multimorbidity and risk of deterioration, with adjustment for inequalities across age, ethnicity and deprivation.
• Geospatial modelling, with indexing against expected levels of co-morbidity, reveals population clusters with highest health inequality, and highest potential for intervention