Table 6.
Summary of characteristic of studies that investigated physical fatigue quantification using wearable devices.
| No. | Type of study | # of participants | Fatiguing task | Task duration | Input data | Reference measure | Modeling approach | Output | Model performance |
|---|---|---|---|---|---|---|---|---|---|
| 45 | Lab | 15/13 | Simulated manual material handling task/Supply pick-up and insertion task | 180 min | Motion, age/ECG | Borg's RPE | Random forest | Binary | Se = 84.7/82.0% Sp = 86.4/88.9% ACC = 85.5/85.4% G-mean = 0.85/0.85 Consistency = 0.15/0.10 |
| 46 | Lab | 8 | Part assembly task, supply pick-up, and insertion task, manual material handling task | 180 min | ECG | Borg's RPE | Random Forest | Binary | ACC = 69.4–90.4% Se = 66.2–82.3% Sp = 71.7–96.2% AUC = 0.86–0.88 |
| 47 | Lab | 9 | 25 m shuttle sprint | Until sprint time decrement of 5% for two consecutive tests | Motion | Not stated | SVM | Binary | ACC = 90% Cohen's kappa = 0.75 |
| 48 | Lab | 12 | Incremental treadmill running test | Time to exhaustion | EMG | Blood lactate samples | Random Forest | 3 states (aerobic, anaerobic and recovery phase) | AUC = 0.86 |
| 49 | Lab | 20 | Treadmill running program, L-drill and step test, crunch and jumps, sit to stand up, and push-up | 7.5 min | Motion | Rate of perceived exertion a day after the protocol | Fatigue score | Binary |
r = 0.95 (male) r = 0.70 (female) |
| 50 | Lab | 6 | Wall building/two bricklaying activities | Approx. 30 min/50 min | Motion | Not stated | SVM with quadratic kernel/SVM with medium gaussian kernel | Binary | ACC = 79.2%/ ACC = 65.9% |
| 51 | Lab | 6 | Lifting/lowering and turning task in 2 different paces (quick/slow) | 5 min each task | RES, GSR, PPG | Borg's RPE | Linear regression model | 3 levels of fatigue | Correct rate = 66.7% Absolute difference = 1.9 R2 = 0.39 |
| 52 | Lab | 15 | Jumping rope consecutively | 5 min (repeated until exhaustion) | Motion, age, height, weight | Maximum rope number | Heterogeneous ensemble learning voting method | Binary | ACC = 92% Precision = recall = = F1-score = 0.73 |
| 53 | Lab | 8 | Simulated manufacturing tasks | 180 min | ECG, motion | Borg's RPE | Least absolute shrinkage and selection operator model | Binary/RPE prediction | Se = 1, Sp = 0.79/ MAE = 2.16 |
| 54 | Lab | 12 | Simulated construction activity | 200 trials (approx. 150 min) | TSk, ECG, personal information | Borg's RPE | Boosted trees | 4 levels | ACC = 82.6% |
| 55 | Lab | 8 | Propel a wheelchair at a constant speed of 1.6 m/s | Until being unable to meet the required speed | ECG, EMG, motion | Self-reported fatigue | Neuro-fuzzy classifier | 3 levels | ACC = 80% |
| 56 | Field | 21 | The Beep test or Pacer test until exhaustion | Time to exhaustion or Borg's RPE ≥18 | Motion | Borg's RPE | Random Forest | Binary | ACC = 75% Se = 73% Sp = 77% F1-score = 75% |
ECG,electroencephalogram; EMG,electromyogram; GSR,galvanic skin response; PPG,photoplethysmogram; RES,respiration; TSk,skin temperature; RPE,ratings of perceived exertion; SVM,Support Vector Machine; ACC,accuracy; AUC,area under the receiver operating characteristic curve; MAE,mean absolute error; r,correlation coefficient; R2, R-squared; Se, sensitivity; Sp, specificity.