Skip to main content

Table 2 Patient characteristics

From: The practical clinical role of machine learning models with different algorithms in predicting prostate cancer local recurrence after radical prostatectomy

Characteristics

Training set nā€‰=ā€‰123

Testing set nā€‰=ā€‰53

P value

Age (years, meanā€‰Ā±ā€‰SD)

69.4ā€‰Ā±ā€‰6.7

70.1ā€‰Ā±ā€‰6.6

0.938

Pre-operative PSA level [ng/mL, median (IQR)]

19.5 (6.5ā€“32.5)

24.8 (7.4ā€“42.2)

0.835

The time between RP and MRI [months, median (IQR)]

10.3 (2.5ā€“18.1)

12.3 (3.6ā€“21.1)

0.896

Follow-up time [months, median (IQR)]

59.0 (35.5ā€“82.5)

46.0 (30.0ā€“61.0)

0.343

Surgical Gleason score [n, (%)]

Ā Ā 

0.816

6

6 (4.9%)

3 (5.7%)

Ā 

7 (3ā€‰+ā€‰4)

29 (23.6%)

9 (17.0%)

Ā 

7 (4ā€‰+ā€‰3)

35 (28.5%)

18 (34.0%)

Ā 

8

19 (15.4%)

10 (18.9%)

Ā 

9ā€“10

34 (27.6%)

13 (24.5%)

Ā 

Pathologic T stage [n, (%)]

Ā Ā 

0.168

T2a/b

8 (6.5%)

9 (17.0%)

Ā 

T2c

42 (34.1%)

16 (30.2%)

Ā 

T3a

35 (28.5%)

12 (22.6%)

Ā 

T3b

19 (15.4%)

11 (20.8%)

Ā 

T4

19 (15.4%)

5 (9.4%)

Ā 

SVI [n, (%)]

29 (23.6%)

13 (24.5%)

0.892

PNI [n, (%)]

60 (48.8%)

19 (35.8%)

0.114

PSM [n, (%)]

59 (48.0%)

29 (54.7%)

0.411

Local recurrence evident [n, (%)]

38 (30.9%)

16 (30.2%)

0.926

Pre-operative PI-RADS score [n, (%)]

Ā Ā 

0.622

1ā€“2

7 (5.7%)

1 (1.9%)

Ā 

3

13 (10.6%)

5 (9.4%)

Ā 

4

34 (27.6%)

13 (24.5%)

Ā 

5

69 (56.1%)

34 (64.2%)

Ā 

Post-operative PI-RR score [n, (%)]

Ā Ā 

0.473

1

76 (61.8%)

31 (58.5%)

Ā 

2

10 (8.1%)

8 (15.1%)

Ā 

3

10 (8.1%)

5 (9.4%)

Ā 

4

15 (12.2%)

3 (5.7%)

Ā 

5

12 (9.8%)

6 (11.3%)

Ā 
  1. SDā€‰=ā€‰standard deviation; IQRā€‰=ā€‰interquartile range; PSAā€‰=ā€‰prostate specific antigen; RPā€‰=ā€‰radical prostatectomy; SVIā€‰=ā€‰seminal vesicle invasion; PNIā€‰=ā€‰perineural invasion; PSMā€‰=ā€‰positive surgical margins; PI-RADSā€‰=ā€‰Prostate Imaging Reporting and Data System; PI-RRā€‰=ā€‰Prostate Imaging for Recurrence Reporting