Risk prediction of chronic diseases with a two-stage semi-supervised clustering method

Figures

• 

  Fig. 1. Overview workflow.

• 

  Fig. 2. Polynomial fitting for age-dependent parameters. (a): the distribution of red blood cell count. (b): the distribution of diastolic blood pressure. (c) the distribution of systolic blood pressure. The median value for each age point is represented as dots. Red for female and blue for male. The fitted quartic polynomial are shown as red and blue lines. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

• 

  Fig. 3. Training procedure.

• 

  Fig. 4. Validation procedure. Left: performance evaluation using training data. Right: performance evaluation using validation data.

• 

  Fig. 5. Prevalence of CVD vs age. Red dashed line, mean value of high-risk group in the training data. Green line, mean value of low-risk group in the training data. Red line, mean value of high-risk group in the validation data. Red band, standard error of high-risk group in the validation data. Green line, mean value of low-risk group in the validation data. Green band, standard error of low-risk group in the validation data. 10% of the labeled data is used as ground truth for SKK-Means clustering. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

• 

  Fig. 6. Diabetes prevalence vs age. Same coloring scheme as Fig. 5.