AI project launched to close heart health gaps in cities

New York is the launch city for a new programme which aims to tackle heart health inequities with the help of artificial intelligence (AI).

Cardiovascular disease is the world’s number one cause of death and disability, killing 17.9 million people each year. In the US, as with many other countries, it disproportionately affects racial and ethnic minorities and people on low incomes.

AI4HealthyCities will bring together data on heart health and social determinants, and apply advanced data analytics to provide insights about the interventions that could have the greatest impact.

The AI4HealthyCities Health Equity Network was created by the Novartis Foundation and developed in collaboration with Microsoft AI for Health. In New York City, the initiative is run in partnership with the New York University School of Global Public Health, the public hospital system NYC Health + Hospitals, and the city health authorities.

Additional cities will be announced soon, with each city programme examining different factors impacting heart health.

Dr Ann Aerts, head of the Novartis Foundation, told Cities Today: “The impact of social determinants of health, which are factors such as food, employment, green space and structural racism, are widely known and accepted. It can be difficult, however, for policymakers and other decision-makers to know exactly which combination of factors have the greatest impact on the health of a given community, and what concrete steps could have the greatest positive impact.

“These are the insights that we aim to deliver, and which we hope can help decision-makers make more informed decisions about how to channel scarce resources in a way that improves the health of entire populations, including decisions about which sectors they need to forge collaborations with.”

Street-level data

The NYU School of Global Public Health recently conducted a literature review and stakeholder interviews to understand what is currently known about the influence of social, economic and environmental determinants on heart health in the city’s population, and what has already been done to remediate inequity. They also made an inventory of publicly available datasets including social determinants of health and obtained sample patient-level data from NYC Health + Hospitals.

The project will use “sophisticated algorithms” to ‘learn’ features from a large volume of data and extract useful information such as patterns and predicted outcomes. This will produce zip code-level data visualisation tools, maps and vulnerability risk scores.

Juan Lavista Ferres, Chief Data Scientist at Microsoft, said: “Combining anonymised population data from many different sources enables us to use machine learning to go down to street level in assessing levels of cardiovascular risk. If properly applied this could revolutionise preventive interventions in heart health.”

Aerts said steps are being taken to mitigate AI-related risks: “Publicly available data will be stored on Microsoft Azure, a secure cloud space, and analysed by our academic partner. The project will be regulated by data use agreements between all parties to specify the period of use of the data, exchanges, and security modalities, and the research protocol will undergo ethical approval whenever necessary.”

She added: “One risk that we are being particularly vigilant against is the biases that we know can impact outputs generated by AI. We will address bias in the AI models by employing a diverse team to work on the models, using large, diverse datasets, and making close and critical evaluations of outputs.”

Data for health

AI4HealthyCities builds on the Novartis Foundation’s CARDIO4Cities approach to reduce cardiovascular risk. In three initial pilot cities – São Paulo, Dakar and Ulaanbaatar – the organisation reports that the initiative tripled blood pressure control rates. The Novartis Foundation is also expanding the approach to include other cardiovascular risk factors, such as high cholesterol, diabetes and obesity, in São Paulo, Dakar and Ho Chi Minh City.

Other cities are also tapping AI to address the social determinants of health. A collaboration between De Montfort University and Leicester City Council in the UK is using various datasets to provide more personalised support to tackle obesity and its related risks.

Dr Ioannis Kypraios, Associate Professor in Artificial Intelligence and co-principal investigator of the project at DMU, told Cities Today earlier this year: “This big data is ideal for machine learning artificial intelligence to be able to understand the connections between the different features and then find these hidden trends that exist across all the different datasets, and use it to empower the practitioners.

“Normally these kinds of data are completely unrelated and now we are bringing them all together.”

Another project in Mexico used AI to improve the detection and prevention of diabetic retinopathy.