Background There has been a renewed interest in health inequality and the social determinants of health; yet most of the focus of the literature is a description of inequality rather than investigating interventions to tackle health inequalities. With an exponential growth in health research, and academic publishing growing at 4% per year: traditional systematic review (SR) methods are becoming increasingly unmanageable, time consuming and quickly outdated. Developments in machine learning have created opportunities for researchers to efficiently navigate large bodies of research. Our aim was to use machine learning (ML) algorithm to produce Living Evidence Maps of what works to address health inequalities in a primary care setting.

Methods Using EPPI Reviewer software,an initial set of six of systematic reviews were used to train a ML algorithm, to identify studies relating to what works to address inequalities in primary care. Additional studies were identified using network graph searches. The ML software prioritised potential studies in OpenAlex according to their similarity to the training material. Prioritised studies were screened on title-abstract and full text, and manually coded for their intervention type, disadvantaged group, and health and care outcomes.

Using EPPI-Visualiser software, the codes for interventions, disadvantaged groups and health and care outcomes were mapped to allow researchers to identify evidence to inform practice, and gaps in research. The Living Evidence Maps are published open access (www.heec.co.uk).

Results Our Living Evidence Map contains a total of 329 SRs and 3 umbrella reviews (UR). Wide variation exists in the quantity of literature exploring different interventions, outcomes, and disadvantaged groups. The five topics with the largest quantity of reviews include: advice and counselling intervention for ethnic minorities (n = 107 SR, n = 2 UR), education intervention for ethnic minorities (n = 106 SR), cultural competence intervention for ethnic minorities (n = 97 SR, n = 1 UR), community and link worker intervention for ethnic minorities (n = 86 SR), and advice and counselling interventions for the socio-economically deprived group (n = 62 SR, n = 2 UR). There was a paucity of evidence on gypsy, roma or traveller community; autism; and ADHD.

Conclusion Our Living Evidence Maps provides a unique, and perpetually up-to-date insight into interventions addressing health inequalities within primary care. The application of the machine learning algorithm enables efficient and effective navigation of literature.