Table 1.
First Author [ref.] |
Year | Cohort | Imaging Modality | ML Algorithm | Main Findings |
---|---|---|---|---|---|
Alzheimer’s Disease and Cognitive Impairment | |||||
Gaser [31] | 2013 | 133 pMCI, 62 sMCI | T1WI | RVR | BAG predicts conversion to AD, 10% greater risk of developing AD by each 1 additional yr of BAG |
Lowe [32] | 2016 | 150 AD, 112 pMCI, 36 sMCI, 107HC | T1WI | RVR | Effect of APOEe4 on BrainAGE changing rates over time |
Beheshti [17] | 2018 | 147 AD, 112 pMCI, 102 sMCI, 146 HCs | T1WI | SVR | BAG: +5.36 yr in AD, +3.15 yr in pMCI, +2.38 yr in sMCI. Correlation with cognitive function |
Wang [33] | 2019 | 3688 people (middle age to elderly) | T1WI | CNN | BAG: related to incident dementia |
Mohajer [35] | 2020 | 48 AD, 222 MCI, 60 HCs | T1WI | SVR | BAG was elevated in MCI and AD but was not associated with sleep-disordered breathing. |
Ly [34] | 2020 | 74 AD, 283 MCI, 51 preclinical AD, 83 HCs | T1WI | GPR | BAG differentiated cognitively unimpaired Amyloid (+) from Amyloid (−). |
Beheshti [23] | 2021 | 292 AD, 440 MCI, 548 HCs | FDG-PET | SVR | Younger BAG in females than in males in HCs group but not in MCI or AD groups |
Habes [36] | 2021 | 1932 MCI/AD, 8284 HCs | T1WI | RBF-kernel | BAG associated with WMH as well as cognitive function |
Parkinson’s disease | |||||
Beheshti [20] | 2020 | 160 PD, 129 AD, 839 HCs | T1WI | SVR | GM-based BAG: +1.50 yr in PD, +9.29 yr in AD. WM-based BAG: +2.47 yr in PD, +8.85 yr in AD. WM-based BAG > GM-based BAG in PD |
Eickhoff [42] | 2021 | 372 PD, 172 HCs | T1WI | SVR | BAG: +2.9 yr in PD. Associated with disease duration and cognitive and motor impairment. |
Charisse [43] | 2022 | 83 PD-NC, 78 PD-MCI, 17 PD-D, 84 HCs | T1WI | SVR | RBA: +2.38 yr in PD-NC, +1.90 yr in PD-MCI, +3.52 yr in PD-D. Associated with attention deficits and working memory |
Epilepsy | |||||
Pardoe [44] | 2017 | 42 new FE, 94 refractory FE, 74 HCs | T1WI | GPR | BAG: +4.5yr in refractory FE, no significance in new FE |
Hwang [45] | 2020 | 104 TLE, 151 HCs | T1WI, fMRI | SVR | T1-based BAG: +6.6 yr in TLE. fMRI-based BAG: +8.3 yr in TLE Association with clinical data |
Sone [19] | 2021 | 318 epilepsy, 1,196 HCs | T1WI | SVR | BAG: >+4 yr in all types of epilepsies, +10.9 yr in TLE with psychosis |
de Bézenac [46] | 2022 | 48 TLE, 37 HCs | T1WI | GPR | BAG: +7.97 yr in TLE, postsurgical reduction of BAG |
Multiple sclerosis | |||||
Cole [47] | 2020 | 1204 MS/CIS, 150 HCs | T1WI | GPR | BAG: +10.3 yr in MS, +13.3 yr in SPMS, predictive value for progression |
Jacobs [48] | 2021 | 179 MS | T1WI | GPR | BAG: +6.54 yr in MS, associated with a physical disability |
Traumatic brain injury | |||||
Gan [49] | 2021 | 116 mTBI, 63 HCs | DTI | RVR | BAG: +2.59 yr in mTBI, associated with post-concussion complaints |
Hellstrom [50] | 2021 | 123 mTBI | T1WI, DTI | XGBoost | No significant difference in BAG between APOEe4 carriers and non-carriers after mTBI |
Pain | |||||
Yu [51] | 2021 | 31 CLBP, 32 HCs | T1WI | GPR | Discrepancy in BAG between HCs and CLBP was greater in older individuals |
Hung [52] | 2022 | 45 TN, 52 OA, 50 CLBP, 812 HCs | T1WI | GPR | BAG: +6.48 yr in TN, +9.80 yr in OA, no significance in BP. Female-driven elevation in BAG |
Others | |||||
Azor [53] | 2019 | 20 PWS, 40 HCs | T1WI | GPR | BAG: +7.24 yr in PWS, Not associated with IQ, hormonal or psychotropic medications, or abnormal behaviors |
Cole [54] | 2017 | 162 HIV(+), 105 HIV(−) | T1WI | GPR | BAG: +2.15 yr in HIV(+), associated with cognitive performance |
AD: Alzheimer’s disease, BAG: brain age gap, CIS: clinically isolated syndrome, CLBP: chronic lower back pain, CNN: convolutional neural network, DTI: diffusion tensor imaging, FDG-PET: 18F-fluorodeoxyglucose PET, FE: focal epilepsy, fMRI: functional MRI, GPR: Gaussian process regression, HCs: healthy controls, ML: machine learning, MS: multiple sclerosis, mTBI: mild traumatic brain injury, NC: normal cognition, OA: osteoarthritis, PD: Parkinson disease, PD-D: PD with dementia, pMCI: progressive mild cognitive impairment, PWS: Prader-Willi syndrome, RBF: radial basis function, RVR: relevance vector regression, sMCI: stable mild cognitive impairment, SPMS: secondary progressive MS, SVR: support vector regression, T1WI: T1-weighted image, TLE: temporal lobe epilepsy, TN: trigeminal neuralgia.