HEALTHBOT.AI: a conversational prototype exploring cardiovascular risk

Editorial note: this article documents an educational project and its results on the data used. It does not establish clinical validation or authorisation to diagnose or treat patients. Proposed healthcare applications are retained as the team’s goals, not demonstrated capabilities.

HealthBot began during the COVID-19 pandemic in response to difficulties accessing healthcare. It proposes a conversational interface to collect information on cardiovascular risk factors and explore how to present that information to users and clinicians. The first prototype builds on an earlier Saturdays.AI project, CardioSight. Possible integration with public healthcare services and expansion to other conditions remain future work requiring validation.

Context and problem

Cardiovascular diseases are a major cause of death worldwide. WHO highlights risk factors such as tobacco use, high blood pressure, unhealthy diets and physical inactivity, alongside social and environmental factors. This context motivated the team to explore support tools. The mixed figures in the original version did not reliably support attributing 77% of adult deaths in Mexico to cardiovascular diseases, so that claim has been removed.
Cardiovascular diseases have been one of the main causes of death in the world. With the figures obtained in 2012, there were 17.5 million deaths registered worldwide. More than 80% of deaths mainly affect the low and middle countries, equally in men and women. In Mexico, 77% of deaths in adults are due to cardiovascular diseases, which is why risk factors are involved, such as 17% of the population being smokers, 22.8% suffering from hypertension and 1 in 3 adults being obese. Of which, obesity has been one of the main risk factors. This is due to the high consumption of sugar, and the lack of education on these issues in the country.

Solution Proposal

HealthBot proposes a text chatbot to collect information and present an experimental estimate of cardiovascular risk. The team planned to include preventive information and generate summaries users could share with their doctors. It also considered possible integration with institutions such as IMSS. These are design goals: the prototype does not establish a diagnosis and should not recommend treatments.

Project development

The chatbot uses natural language processing (NLP) to collect data through conversation. Its reference model comes from CardioSight, developed by the orange team in the previous Saturdays.AI cohort. Reusing that work allows this project to focus on the conversational interface and explore ways to present results, without assuming the model is validated for clinical use.

Cardiovascular disease risk prediction model

We use the orange team’s model as a reference. The variables described include age, gender, systolic and diastolic blood pressure, cholesterol, glucose, physical activity, height in centimetres, weight in kilograms and body mass index. The figure shows correlations between the variables after that team’s exploratory analysis.

Figure 1. Result of the correlation of the aforementioned variables

Technical Description of the Project

Credits

We thank our mentors, Grisell Reyes Rios and Mario Fonseca, for their support and guidance throughout the project.

More artificial intelligence!

Saturdays.AI makes AI more accessible through shared learning and projects with a social purpose. Our motto is AI4ALL IS AI4GOOD.