Machine Learning Algorithms Identify Top Predictors of Noncalcified Coronary Burden in Psoriasis

AI Artificial intelligence machine learning
Machine learning methods identified top predictors of coronary artery burden in patients with psoriasis, which were markers related to obesity, dyslipidemia, and inflammation, demonstrating that these are potentially important comorbidities to treat in psoriasis.

Markers of obesity, dyslipidemia, and inflammation were found by coronary computed tomography angiography (CCTA) to represent top predictors of noncalcified coronary burden in patients with psoriasis. The markers were identified by machine learning algorithms and detailed in a study published in the Journal of the American Academy of Dermatology.

A total of 263 consecutive patient records in which 62 phenotypic variables were present at baseline were obtained from the ongoing Psoriasis Atherosclerosis Cardiometabolic Initiative trial. To identify top predictors of noncalcified coronary burden by CCTA, the researchers used the random forest algorithm. Importance of the 62 variables was validated by permutation within a random forest algorithm comprising 200 regression trees. The CCTA scans were read by investigators blinded to patient characteristics, visit date, and treatment.

Ranking in order from the highest importance to lowest importance, the top predictors of noncalcified coronary burden using the random forest algorithm were body mass index (importance, 0.66), visceral adiposity (0.64), total adiposity (0.41), apolipoprotein A1 level (0.22), high-density lipoprotein level (0.19), erythrocyte sedimentation rate (0.17), subcutaneous adiposity (0.15), small low-density lipoprotein particle (0.13), and cholesterol efflux capacity (0.11).

Related Articles

Variables that featured statistically significant negative associations with noncalcified coronary burden included apolipoprotein A1 level (P <.001), high-density lipoprotein level (P <.001), cholesterol efflux capacity (P <.001), and large medium high-density lipoprotein particle (P =.01).

Limitations of the study were the inclusion of only baseline values from the patients’ first visit as well as the inclusion of only patients from the Psoriasis Atherosclerosis Cardiometabolic Initiative.

“Further investigation into these top predictors of noncalcified coronary burden over time may provide insight into the treatment of inflammation and comorbidities in psoriasis to reduce cardiovascular disease risk,” the study authors wrote.

Disclosure: Nehal N. Mehta, MD, and Joel M. Gelfand, MD, declared affiliations with the pharmaceutical industry. Please see the original reference for a full list of authors’ disclosures.


Munger E, Choi H, Dey AK, et al. Application of machine learning to determine top predictors of non-calcified coronary burden in psoriasis: an observational cohort study [published online October 30, 2019]. J Am Acad Dermatol. doi:10.1016/j.jaad.2019.10.060

This article originally appeared on Dermatology Advisor