Skip to main content

Table 4 Characteristics of the models

From: Prognostic models for breast cancer: a systematic review

 

Number of modelsa

Total

58 models

Types of models

 New models

49 models

 Modified models

9 models

Year of development

1982–2016

 1982–1989

5 models

 1990–1999

11 models

 2000–2009

17 models

 2010–2016

25 models

Country of participants for model development

 Europe

25 models

 Asia

13 models

 North America

12 models

 Others

1 model (Australia)

 Unknown or from several trials

7 models

Method of model development

 Cox PH regression

32 models

 Artificial neural networks

6 models

 Decision tree

4 models

 Logistic regression

3 models

 Bayesian method

3 models

 Multistate model

2 models

 Support vector machine

2 models

 Others

6 models

Outcomes

 Mortality

28 models

 Recurrence

23 models

 Both

7 models

Predictors

 Age at diagnosis

24 models

 Nodal status

49 models

 Tumour size

42 models

 Tumour grade

29 models

 Lympho-vascular invasion (LVI)

8 models

 Stage

8 models

 ER status

21 models

 Progesterone receptor (PR) status

10 models

 HER2 status

13 models

 Treatment

17 models

 Others

Mitotic activity index (MAI), histological subtypes, comorbidity, menopausal status, etc.

Presentation of model

 Regression formula

13 models

 Online tool

8 models

 Decision tree

5 models

 Nomogram

4 models

 Score chart

1 model

 No report

27 models

Number of risk groups

 5

3 models

 4

3 models

 3

9 models

 2

6 models

 No report/No risk group

33 models

Validation

 No validation

11 models

 Internal validation

43 models

 External validation

17 models

  1. aTotal number of models is 58. Where each model can fit more than one category, the number of models may not always total 58