ANA MARIA LUCAS OCHOA¹, Sandra Amador Moreno², Emiliano Fernández Villalba², Lorena Cuenca Bermejo¹, José Luis Fernández Alemán³, Ana María González Cuello², David Gil⁴, Valeria Raparelli⁵, Colleen Norris⁶, Alexandra Kautzky-Willer⁷, Karolina Kublickiene⁸, Louise Pilote⁹, Maria Trinidad Herrero Ezquerro¹

(1) NEUROCIENCIA CLÍNICA Y EXPERIMENTAL, IMIB-Arrixaca, España
(2) Clinical and Experimental Neuroscience (NiCE), Institute for Aging Research, Biomedical Institute for Bio-Health Research of Murcia (IMIB-Arrixaca), School of Medicine, Campus Mare Nostrum, University of Murcia, España (Región de Murcia)
(3) Department of Informatics and Systems, Faculty of Computer Science, University of Murcia, España (Región de Murcia)
(4) Lucentia Research Group, Department of Computer Science Technology and Computation, University of Alicante, España (Comunidad Valenciana)
(5) Department of Transnational Medicine, University of Ferrara, Ferrara, Italia
(6) Faculty of Nursing, University of Alberta, Edmonton, Alta, Cardiovascular and Stroke Strategic Clinical Network, Alberta Health Services, Edmonton, Alta, Canadá
(7) Division of Endocrinology and Metabolism, Department of Medicine III, Medical University of Vienna, Vienna, Austria
(8) Department of Renal Medicine, Institution for Clinical Science, Intervention and Technology, Karolinska Institute, Stockholm, Suecia
(9) Research Institute of McGill University Health Centre, Division of Clinical Epidemiology McGill University, Montreal, Canadá

Cardiovascular diseases (CVDs) remain to be the most frequent cause of mortality in human population worldwide. Importantly, CVDs are the ones who have a higher prevalence of deaths in women. Sex, gender and social factors are key variables that determine differences among individuals in both health and disease. However, few studies include them in the experimental design and, therefore, we could be losing critical information for interpretation, validation, reproducibility and generalizability of research findings. Due to that, the aim of this mapping study was to identify and analyse the articles that include ML techniques, gender, sex and social factors (economic, cultural, demographic and psychological) in the circulatory system diseases. In the last years, the application of machine learning techniques has become a useful approach for solving problems such as help clinicians to make more accurate predictions and improve estimated CVD risk scores to automate predictions.

A mapping study was applied to identify and analyse the articles that applied machine learning techniques to the CVDs considering gender, sex and social factors to answer the research questions. An exhaustive search chain was performed to classify the articles, using the International Statistical Classification of Diseases and Related Health Problems 10th revision (ICD-10), 2019 version, in group IX (Diseases of the Circulatory system) in Scopus database.

We found that only 24 out of more than 2 million studies of the CVDs considered machine learning, social factors, gender and sex. Among them, hypertension and heart failure were the most studied conditions. If the search was performed individually (each variable independently), for machine learning techniques heart disease was the most studied while for gender and social factors was hypertension. Noteworthy, the number of works has gradually increased, especially in the last three years. United States was the country with the higher number of studies published. Decision Tree and Logistic Regression were the machine learning algorithms most employed in the cardiovascular diseases. This requires considerable further reflection and discussion in the light of the number of methods available.

It is important to study social factors in cardiovascular diseases studies with a sex and gender perspective. As aforementioned, it is certainly that an increase in publications has been produced considerably in recent years and the trend is for it to continue to do so in the immediate future. In fact, more research applying machine learning in the diseases of the circulatory system with a sex, gender and social perspective are needed.