A systematic review and meta-analysis of Machine Learning Models for early detection of Coronary Heart Disease Using Nutritional data

Background and aims: According to Who statistics, heart disease is the first cause of mortalityin the world. Coronary heart disease (CHD) is the most common type of heart disease, Nearly۷.۲% of adults have this disease. The diagnosis of this disease usually happens late and when thedisease reaches its acute stages and complications such as appear pain and discomfort in the chest.Therefore, early diagnosis of this disease can greatly reduce the occurrence of problems such asheart attack and death. Today, there are solutions based on diagnosis and prevention using dietarydata and artificial intelligence that can do this in a short period of time.Method: Keywords “diet”, “nutritional science”, “coronary heart diseases”, “artificial intelligence”and “pattern recognition” was used for a comprehensive systematic search in ۳ databasesPubMed, Scopus, Web of Science and the results were up to March ۲۰۲۳ were considered. Tworeviewers reviewed the results independently and separately. Studies that used methods otherthan artificial intelligence to diagnose and prevent coronary artery disease were excluded. Finally,studies that met the necessary inclusion criteria were critically appraised by two authors separately.“Rayyan” platform was used for screening and Microsoft Excel ۲۰۱۹ software was used toextract data related to diet and artificial intelligence diagnostic solutions.Results: At first, ۴۳۷ related publications collected from online databases were retrieved, screeningof titles and abstracts was performed, and duplicate publications(n=۴۹), and ۳۸۸ were removed.The full texts of ۷۷ articles were reviewed. Finally, the studies that met the desired inclusioncriteria were included in the ۶ studies.The algorithm “Logistic regression” was the most used in the field of diagnosis of coronary arterydiseases related to nutrition and it was used in ۴ studies. The total number of data included in thestudies from healthy patients was ۳۷۰۷۰. The ultimate accuracy which was obtained from thesestudies was ۰.۸۸۲.Conclusion: Since heart diseases have a significant impact on the economic conditions of thecountry, the use of artificial intelligence methods, which is a non-invasive method, is a more practicalsolution and separates many nutritional factors and starts diagnosis and prevention earlier.But the accuracy of the studies shows that more work should be done in this field to reach thedesired accuracy for high-certainty diagnosis.

کلیدواژه ها:

Coronary Heart Diseases (CHD) ، Artificial Intelligence ، Nutritional Sciences ، detection