Technology helps medicine predict the risk of disease, but it must take into account the social factors that characterize people’s lives. At least when it comes to cardiovascular disease. This is what a US research group has shown by analyzing the machine learning algorithms applied to risk prediction and best treatment for different groups of patients with cardiovascular diseases. The need to feed the computing power of machines not only with biological but also economic, social and behavioral parameters is very strong in a field where lifestyles and habits make the difference: cardiovascular diseases, in fact, they mainly affect people in disadvantaged socio-economic situations.
“Cardiovascular disease is on the rise, especially in low- and middle-income countries and among black communities in countries like the United States,” said Rumi Chunara, associate professor of biostatistics at New York University’s School of Global Public Health and of computer science. science and engineering at the Tandon School of Engineering, as well as one of the authors of the study published in the American Journal of Preventive Medicine. “The speed with which social and environmental factors have changed in recent decades, such as the amount of processed food present in ‘nutrition, is becoming preponderant with respect to genetic factors whose influence is seen only in the long term ”.
Machine learning algorithms – a type of artificial intelligence that “learns” to interpret data after being fed with information sets – are at the heart of numerous studies on cardiovascular disease. Predictive models are calculated on the basis of clinical information, such as blood pressure and cholesterol levels, but so-called social determinants, such as the neighborhood of residence, are also rarely included. But it is precisely in these cases that machine learning could be more useful because environmental and social factors have complex and non-linear interactions with cardiovascular diseases, which are therefore difficult to interpret. As the study showed after analyzing more than 1600 articles published in scientific literature between 1995 and 2020: the inclusion of social determinants in algorithms improves the ability to predict certain events such as re-hospitalization, heart attack and stroke.
Unfortunately, the authors point out, the models almost never include the complete list of social and environmental variables at the individual community level: some take into account marital status, pay, degree of social isolation, pollution and the presence of insurance. health but only 5 also consider the possibility of walking around their neighborhood and the presence of food shops near the home. Finally, there is a lack of data on populations particularly interested in the increase in the incidence of these diseases, such as those of South America, Africa or Southeast Asia. “Including the social determinants of health can help us understand where the inequalities are and where corrective action is needed,” concludes Chunara. “For example, it can help doctors in their clinical practice identify patients who need social support. -economic to report them to social services, reinforcing the synergy between the health of individuals and that of communities “.
.