| Hayes et al. [26] |
Observational cross-sectional study Healthy (n = 7; mean age: 90.0 y; F:M = 5:2); MCI (n = 7; mean age: 88.4 y; F:M = 4:3) Passive infrared motion sensors and magnetic contact door sensors Six months
|
- Walking speed was more variable in patients with MCI.
- Day-to-day pattern of activities was more variable in patients with MCI. |
| Dodge et al. [27] |
Observational longitudinal study Healthy (n = 54; mean age: 84.9 y; %F: 91%); amnestic MCI (n = 8; mean age: 84.5 y; %F: 88%); non-amnestic MCI (n = 31; mean age: 83.8 y; %F: 84%) Passive infrared sensors Three years
|
- Daily walking speeds and their variability were associated with non-amnestic MCI. |
| Hayes et al. [28] |
Observational cross-sectional study Healthy (n = 29; mean age: 87.5 y; F:M = 26:3); amnestic MCI (n = 6; mean age: 84.8 y; F:M = 5:1); non-amnestic MCI (n = 10; mean age: 86.5 y; F:M = 9:1) Wireless passive infrared motion sensors and magnetic contact door sensors Six months
|
- Patients with amnestic MCI showed less sleep disturbance than both those with non-amnestic MCI and healthy elderly. |
| Petersen et al. [29] |
Observational study Healthy (n = 75; mean age: not clear; F:M = not clear); MCI (n = 10; mean age: not clear; F:M = not clear) Pyroelectric infrared motion sensors and contact sensors One year
|
- Patients with MCI spent an average 1.67 h more inside the home than healthy elderly. |
| Urwyler et al. [30] |
Observational study Healthy (n = 10; mean age: 73.9 y; F:M = 7:3); dementia (n = 10; mean age: 76.7 y; F:M = 7:3) A wireless-unobtrusive sensors (temperature, humidity, luminescence, presence [passive infrared radiation], and acceleration) Twenty consecutive days
|
- Patients with dementia showed unorganized behavior patterns. |
| Rawtaer et al. [31] |
Observational cross-sectional study Healthy (n = 21; mean age: 73.0 y; F:M = 14:7); MCI (n = 28; mean age: 75.1 y; F:M = 19:9) Multiple sensor system (passive infrared motion sensors, proximity beacon tags, a sensor equipped medication box, a bed sensor, and a wearable sensor) Two months
|
- Patients with MCI were less active than healthy subjects and had more sleep interruptions per night.
- Patients with MCI had forgotten their medications more times per month than healthy subjects. |
| Akl et al. [34] |
Observational longitudinal study Healthy (n = 79; mean age: not clear; F:M = not clear); MCI (n = 18; mean age: not clear; F:M = not clear) Passive infrared motion sensors and wireless contact switches Three years Support vector machine, random forest
|
- Variabilities in weekly walking speed, morning and evening walking speeds, and subjects’ age and gender were the most important for the process of detecting MCI.
- This study autonomously detected MCI with receiver operating characteristic curve (0.97) and precision–recall curve (0.93) using a time windows of 24 weeks. |
| Akl et al. [35] |
Observational longitudinal study Healthy (n = 59; mean age: not clear; F:M = not clear); amnestic MCI (n = 11; mean age: not clear; F:M = not clear); non-amnestic MCI (n = 15; mean age: not clear; F:M = not clear) Passive infrared motion sensors and wireless contact switches Three years Clustering (affinity propagation)
|
- This study automatically detected MCI (F0.5 score, 0.856) and non-amnestic MCI (F0.5 score, 0.958). |
| Alberdi et al. [36] |
Observational longitudinal study Healthy (n = 13; mean age: 82.85 y; F:M = 9:4); at risk (n = 10; mean age: 86.20 y; F:M = 10:3); MCI (n = 6; mean age: 84.50 y; F:M = 5:1) Passive infrared motion sensors Two years Regression: support vector regression, linear regression, K nearest neighbors; Classification: support vector machine, adaboost, multilayer perceptron, random forest
|
- Sleep and overnight patterns along with daily routine features contributed to the prediction of several health assessments.
- All algorithms could build statistically significant prediction models. |
| Nakaoku et al. [37] |
Observational study Normal cognition (n = 55; mean age: 75.0 y; F:M = 18:37); cognitive impairment (n = 23; mean age: 78.0 y; F:M = 6:17) Unobtrusive in-house power monitoring system (air conditioner, microwave oven, washing machine, rice cooker, television, and induction heater) One year Generalized linear model
|
- Three independent power monitoring parameters (air conditioner, microwave oven, and induction heater) representing activity behavior were associated with cognitive impairment.
- The prediction model with power monitoring data had better predictive ability (accuracy, 0.82; sensitivity, 0.48; and specificity, 0.96). |
