Table 1 Study characteristics
Author (Year) | Study Design (Level of Evidence) | MINORS SCOREÂ (/16) | Purpose | Source data if external validation study | Database/Institution | TRIPOD Followed? | Number of patients/knees | Mean age at surgery | Female (%) | Follow-up of Outcome (years) |
|---|---|---|---|---|---|---|---|---|---|---|
Martin (2023) | Retrospective Cohort (III) | 12 | Assess sample size effect on accuracy | NR | NKLR + DKLR | Yes | 62,955 | MEDIAN 26 (IQR: 20–36) [Missing data 1870] | 26,446 | 7.6 (4.5) |
Martin (2022a) | Retrospective Cohort (III) | 12 | External Validation of Cox Lasso | NKLR | DKLR | Yes | 10,922 | 29 (11) | 4916 | 8.4 (4.3) |
Martin (2022b) | Case-control (III) | 12 | Determine if machine learning analysis of NKLR can identify the most important risk factors associated with subsequent revision of primary ACL reconstruction | NA | NKLR | NR | 24,935 | 28 (11) | 10,916 | 1, 2, 5 years |
Johnson (2023) | Retrospective Cohort (III) | 12 | Machine learning to predict ACL re-operation | NA | Rochester Epidemiology Project | Yes | 1400 | 27 | NR | 9 years (min 2 years) |
Lopez (2023) | Retrospective Comparative Prognostic (IV) | 11 | Machine learning (ML) models to predict outcomes following ACLR | NR | American College of Surgeons National Surgical Quality Improvement Program database | NR | 21,636 | 31.8 (10.5) | 7638 (35.3%) | 30 days |
Ye (2022) | Case-control (III) | 12 | Machine learning to determine objective and subjective clinical outcomes of ACLR and to determine the most important predictors | NA | Shanghai Sixth People’s Hospital | NR | 432 | 26.8 (8.4) | 112 (25.9%) | 6 years (3.1) |
Martin (2024) | Retrospective Cohort (III) | 11 | To assess the external validity of the NKLR model using STABILITY 1 RCT | NKLR | NKLR + STABILITY 1 trial | NR | 591 | 19.0 (3.2) | 304 (51.4) | 1, 2 |
Jurgensmeier (2023) | Retrospective Cohort (III) | 11 | Machine learning to determine risk of secondary meniscal injury post primary ACLR | NA | Rochester Epidemiology Project (REP) | Yes | 1187 | 25 (18–34) | 502 (42.3%) | 12.3 (6.6–17.6) |
Lu (2022) | Retrospective Cohort (III) | 12 | Machine learning to compare risk and timing of secondary meniscal injury between nonoperative, delayed ACLR, and early ACLR patients | NA | Rochester Epidemiology Project (REP) | Yes | 1369 | 28 (18–37) | 677 (40.7) | min 2 year |