Table 1 Patient Demographics and Clinical Characteristics by Center

From: Enhancing privacy-preserving deployable large language models for perioperative complication detection: a targeted strategy with LoRA fine-tuning

Characteristic

Center 1

Center 2

Number of patients

146

102

Gender, n (%)

  

Male

96 (65.8%)

62 (60.8%)

Female

50 (34.2%)

40 (39.2%)

Age, years (Mean ± SD)

57.7 ± 14.3

59 ± 13.7

Surgical specialty, n (%)

  

Head and neck/other

4 (2.7%)

8 (7.8%)

Gynecological

5 (3.4%)

3 (2.9%)

Cardiovascular

5 (3.4%)

2 (2.0%)

Urological

16 (11.0%)

14 (13.70%)

Neurosurgery

3 (2.1%)

3 (2.9%)

Hepatobiliary and pancreatic

48 (32.9%)

17 (16.7%)

Gastrointestinal

65 (44.5%)

21 (20.6%)

Thoracic

0 (0.0%)

2 (2%)

Orthopedic

0 (0.0%)

15 (14.7%)

Median Length of hospital stay, days (IQR)

12 (9-15)

10 (7-15)

Complications, n (%)

  

Paralytic ileus

36 (24.7%)

20 (19.6%)

Organ/space surgical site infection

21 (14.4%)

16 (15.7%)

Infection of unknown source

22 (15.1%)

3 (2.9%)

Myocardial injury after non-cardiac surgery

20 (13.7%)

7 (6.9%)

Anastomotic leakage

15 (10.3%)

10 (9.8%)

Pneumonia

10 (6.8%)

17 (6.7%)

Postoperative bleeding

10 (6.8%)

15 (14.7%)

Acute kidney injury

7 (4.8%)

12 (11.8%)

Superficial surgical site infection

4 (2.7%)

8 (7.8%)

Arrhythmia

7 (4.8%)

6 (5.9%)

Cardiogenic pulmonary edema

7 (4.8%)

1 (1.0%)

Deep vein thrombosis

5 (3.4%)

9 (8.8%)

Deep surgical site infection

3 (2.1%)

7 (6.9%)

Gastrointestinal bleeding

4 (2.7%)

6 (5.9%)

Delirium

5 (3.4%)

0 (0.0%)

Acute respiratory distress syndrome

4 (2.7%)

2 (2.0%)

Pulmonary embolism

3 (2.1%)

1 (1.0%)

Myocardial infarction

2 (1.4%)

1 (1.0%)

Bloodstream infection

1 (0.7%)

2 (2.0%)

Stroke

0 (0.0%)

1 (1.0%)