Integrating ultrasound and clinical risk factors to predict carotid plaque vulnerability in gout patients: a machine learning approach

Figures

• 

  FIGURE 1. Flowchart of patients.

• 

  FIGURE 2. Step-by-step flow chart to classify carotid plaques into the different Plaque-RADS categories (13).

• 

  FIGURE 3. Representative ultrasound images of Plaque-RADS Category 3 subtypes (3A–3C). (a,b) RADS 3A plaque at the carotid bifurcation with a maximum wall thickness (MWT) greater than 3 mm. The plaque exhibits a uniform isoechoic appearance (arrow) and features a thick fibrous cap. (a) Gray-scale ultrasound imaging. (b) Doppler flow imaging. (c,d) RADS 3B plaque at the carotid bifurcation with a MWT greater than 3 mm. The plaque contains multiple very low echogenic areas, with most regions lacking a visible (thin) fibrous cap. (c) Gray-scale ultrasound imaging. (d) Doppler flow imaging. (e,f) RADS 3C plaque with a mixed hyperechoic and hypoechoic plaque at the carotid bifurcation, demonstrating ulceration (*) on both two-dimensional and Doppler imaging.

• 

  FIGURE 4. Representative ultrasound features of monosodium urate crystal deposits and associated inflammation. (a) Tophi observed on the lateral aspect of the ankle joint. (b) Double-track sign on the medial side of the ankle joint. (c) Scattered hyperechoic aggregates with bone erosion observed in the 1st metatarsophalangeal joint. (d) Intra-articular blood flow signals detected in the ankle joint capsule, suggesting synovitis.

• 

  FIGURE 5. Multivariable ordinal logistic regression model assessing carotid artery plaque vulnerability in a gout cohort. OR, odds ratio; CI, confidence interval; GFR, glomerular filtration rate; PD, power Doppler. *P < 0.05; **P < 0.01.

• 

  FIGURE 6. Cumulative ROC curves of the two models across different classification thresholds. Model 1 included traditional cardiovascular risk factors: age, cholesterol levels, GFR < 60 mL/min, diabetes, and antihypertensive medication use. Model 2 extended Model 1 by integrating gout-specific factors, including the presence of tophi, Doppler flow, and the number of gout flares in the last year. ROC, receiver operating characteristic; TPR, true positive rate; FPR, false positive rate.

• 

  FIGURE 7. Mean decrease in Gini for feature importance in a random forest model predicting carotid plaque. GFR, glomerular filtration rate; PD, power Doppler.