1 Introduction
1.1 What is ADHD and How is It Currently Diagnosed?
1.2 Movement Sensors in ADHD Assessment
1.3 The Present Study
2 Methods
2.1 Study Selection
2.2 Study Participants
2.3 Sensor Data Collection
2.4 Experimental Designs
2.5 Analysis Methods
3 Results
Author(s) and Publication Year | Number and Gender Ratio of Subjects and Controls | Age range or mean (M) | Sensor Type, Placement and Duration | Analysis Method | Discrimination Accuracy / Effect size | Interpretation |
---|---|---|---|---|---|---|
Studies with Naturalistic Design: | ||||||
Lin et al. 2020 [28] | 15 ADHD, 15 controls (total 27% females) | M: 6.83 SD: 1 | Smartwatch IMUs on wrist 2 h for 3 consecutive days | Two sample t-test, ROC | AUC: 82.0% p < 0.001 | Method is moderately accurate |
Kam et al. 2010 [27] | 10 ADHD, 132 controls | M: 7.44 SD: 0.62 | Actigraph on non-dominant wrist Single 3h period | Decision tree | ACC: 99.3% SE: 100.0% SP: 99.2% | Method is extremely accurate |
Langevin R et al. 2012 [29] | 5 ADHD, 5 medicated ADHD, 5 healthy controls (total 13% females) | M: 8.13 | Actigraph on non-dominant wrist Two periods of 5 consecutive days | Kruskal-Wallis test | Nocturnal movement, group difference by period: p = 0.008 | Significant group differences |
Lindhiem et al. 2022 [30] | 15 ADHD, 15 healthy controls | 6–11 yr | Smartwatch IMUs on wrist Single period of 2 days | SVM, Logistic regression, Random Forest | ACC: 89.0% SE: 93.0% SP: 86.0% | Method is highly accurate |
Gruber R et al. 2011 [31] | 11 ADHD (36.4% female), 32 healthy controls (37.5% female) | M: 8.70 SD: 1.30 | Actigraphy Period of 6 consecutive nights | ANOVA, MANOVA | sleep efficiency: F(1, 38) = 86.18 p < 0.001 | Significant group differences |
Munoz-Organero et al. 2018 [13] | 11 ADHD, 11 healthy controls (total 9% females) | 6–12 yr | Accelometer on dominant wrist and ankle Single 24h period | CNN | ACC: 93.8% SE: 80.0% SP: 100.0% | Method is very accurate |
Licht CA et al. 2009 [32] | 9 ADHD (11% females), 9 healthy controls (22% females) | M: 9.22 SD: 1.09 | Actigraph around the waist 24h for a full 7-day week | ANOVA | Group X Period quadratic effect: F(1, 16) = 5.12, p = 0.04 | Significant group differences |
Tsujii N et al. 2009 [33] | 18 ADHD, 10 PDD, 18 Controls | M: 9.23 SD: 1.45 | Actigraph on non-dominant wrist Single 1 week period | ANOVA, post-hoc Scheffét-test | Average and SD Activity during recess: F = 8.84 and F = 12.11 p < 0.001 | Significant group differences |
Amado-Caballero et al. 2020 [26] | 73 ADHD, 75 healthy controls | 6–15 yr | Actigraph on dominant wrist Single 24 h period | CNN | ACC: 98.6% SE: 97.6% SP: 99.5% | Method is extremely accurate |
Brandt et al. 2012 [34] | 5251 ADHD | Initial: M: 7.00, Follow-up: M: 14.00 | Actigraph around the hip 7 consecutive days during waking hours | Linear regression models | Model 2: F = 477.07 p = 0.001 | Significant group differences |
Faedda et al. 2016 [35] | 44 ADHD, 42 healthy controls, 48 bipolar subjects | 5–18 yr | Actigraph around the waist Continuous period of 3–5 days | ANCOVA | ADHD vs Control Activity levels: p < 10–6 | Significant group differences |
Studies with Laboratory Task: | ||||||
Miyahara et al. 2014 [25] | 93 ADHD (26% females), 76 healthy controls (33% females) | M: 3.72 | Actigraph on waist and non-dominant ankle 2 visits of approx. 2 h each | Discriminant analysis | ACC: 69.8% p < 0.001 | Satisfactory accuracy and significant group differences |
Bhattacharyya et al. 2022 [36] | 10 ADHD, 20 controls | 3–5 yr | EEG on forehead and CCD Single period of 14 min | two-tailed t-test after Welch's correction | Hyperactivity index: t = 8.836, p < 0.0001 | Significant group differences |
Chang et al. 2023 [37] | 31 ADHD, 31 controls (48% females for both groups) | M: 7.66 SD: 2.33 | Smart chair Routine visit, approx. 7–15 min | KNN, SVM, XGBoost | ACC: 92.3% AUC: 98.0% | Method is highly accurate |
O’Mahony et al. 2014 [21] | 24 ADHD (29% females), 19 controls (53% females) | M: 9.00 SD: 1.37 | IMUs on waist and dominant ankle Lab visit, approx. 1h | SVM | ACC: 95.1% SE: 94.44% SP: 95.65% | Method is very accurate |
Rapport MD et al. 2009 [38] | 12 ADHD, 11 Controls (no female) | M: 9.04 SD: 1.36 | Actigraph on non-dominant wrist and both ankles Single 2.5 h period with 2–15 min breaks | ANOVA | Total activity by group: F = 36.55, p < 0.001 | Significant group differences |
Dane AV et al. 2000 [39] | 20 inattentive ADHD (15% female), 22 combined ADHD (18% female), 22 controls (36% female) | M: 9.17 SD: 1.40 | Actigraphy Two 2h periods | Univariate analysis and ANOVA | ADHD vs. Control, by Session: F(1,61) = 8.32 p < 0.01 | Significant group differences |
Inoue K et al. 1998 [40] | 20 ADHD 52 Controls (no female) | M: 9.36 | Actigraph on waist Single 1–2 h period | First 10 min mean activity: SE: 75.0% SP: 62.0% p < 0.01 | Significant group differences | |
Halperin JM et al. 1992 [41] | 31 ADHD, 53 non-ADHD, 18 controls | M: 9.65 SD: 1.82 | Actigraph on waist Single 1h period | ANOVA, ANCOVA | Activity level F = 8.25 p = 0.001 | Significant group differences |
Seesjärvi et al. 2022 [24] | 38 ADHD (13% females), 38 healthy controls (21% females) | M: 10.54 SD: 1.08 | EPELI VR simulation and controllers Total duration max 35 min | Discriminant analysis | Controller motion: AUC: 73.0% SE: 71.0% SP: 66.0% | Significant group differences |
Merzon et al. 2022 [23] | 37 ADHD (22% females), 36 healthy controls (42% females) | M: 10.59 SD: 1.08 | Eye tracker in EPELI VR simulation Single period of 25–35 min | SVM | Controller motion: AUC: 70.0% p = 0.0085 | Significant group differences |
Wood AC et al. 2009 [42] | 116 ADHD (10% female), 119 siblings of ADHD (51% female), 218 controls (22% female) | M: 11.87 SD: 2.62 | Actigraph on waist and dominant leg Single 2h period with 25 min break | t-test, ROC | Leg & waist movement intensity: AUC: 79.0% | Method is moderately accurate |
Halperin JM et al. 2008 [43] | 98 ADHD, 85 controls (no female) | Initial M: 9.09 SD: 1.30 Follow-up: M: 18.40 SD: 1.63 | Actigraph on non-dominant ankle and waist Interview and test battery | MANOVA | Ankle Activity, Cohen’s d: 0.66 p < 0.001 | Significant group differences |
Delcour Jensen et al. 2021 [22] | 10 ADHD 10 controls | Na (Adult) | VR-CPT system and controllers (head, leg, hand) | Two sample t-test | Overall activity: t = 2.33, p = 0.039 | Significant group differences |
Edebol et al. 2013 [44] | 55 ADHD (54% female), 202 healthy controls (44% female), 84 ADHD-normative (44% female) | M: 33.16 SD: 9.82 | QbTest-plus Single 20 min period | Fisher’s exact test | SE: 86.0% SP: 83.0% | Method is moderately accurate |