Table 1 Methodology of included systematic reviews and meta-analyses

Pang et al 2019 [17]

N = 9 (2010–2015)

CINAHL

Embase

MEDLINE

Compendex

To summarise and critically examine evidence regarding the detection of near falls (slips, trips, stumbles, missteps, incorrect weight transfer, or temporary loss of balance) using wearable devices.

Adults (aged > = 18 years of age)

Average per study = 21 participants

Total = 192 participants

N = 3 (accelerometer)

N = 4 (accelerometer and gyroscope)

N = 1 (accelerometer and an Android mobile phone)

N = 1 (multiple sensors)

N = 5 (Accuracy/sensitivity and specificity of 97% or greater)

N = 3 (Accuracy was improved by increasing the number of wearable devices)

N = 2 (Chest and right thigh most accurate location for single device placement)

Nguyen et al 2018 [13]

N = 24 (2015–2017)

Springerlink

Elsevier

IEE Xplore Digital Library

Multidisciplinary Digital Publishing Institute (MDPI)

To systematically evaluate the use of Internet of Things (IoT) technology, especially in terms of sensing techniques and data processing techniques in performing falls management for supporting older adults to live independently and safely.

Adults (aged > = 18 years of age)

Average per study = 7 participants

Total = 170 participants

N = 5 (accelerometer)

N = 2 (accelerometer and gyroscope)

N = 3 (smartphone)

N = 6 (camera or laser)

N = 2 (“wearable sensor”)

N = 3 (multiple devices)

N = 2 (wireless networks)

Wearable devices are effective for falls detection - achieving high specificity, sensitivity, and accuracy. Heterogenous methodology in the included studies make quantitative interpretation difficult.

Montesinos et al 2018 [10]

N = 13 (2008–2014)

PubMed

Embase

IEEE Xplore

Cochrane Central Registry of Controlled Trials (CENTRAL)

ClinicalTrials.gov

World Health Organisation International Clinical Trials Registry Platform

To synthetize the empirical evidence regarding inertial sensor-based falls risk assessment and prediction to identify optimal combination of sensor placement, task and features aiming to support evidence-based design of new studies and real-life applications.

At least 10 participants with an average age of 60 years old or over with no severe cognitive or motor impairment.

Studies in which participants were labelled as fallers and non-fallers.

Average per study = 93

Total = 1211 participants

N = 9 (accelerometer)

N = 3 (accelerometer and gyroscope)

N = 1 (gyroscope)

The statistical analysis of features reported in the 13 shortlisted studies revealed significant, very strong, positive associations in 3 different triads of feature category, task, and sensor placement:

• Angular velocity – Walking – Shins

• Linear acceleration – Quiet standing – Lower back

• Linear acceleration – Stand to sit/Sit to stand – Lower back

Chaudhuri et al 2014 [16]

N = 57 (2007–2013)

PubMed

CINAHL

Embase

PsycINFO

To systematically assess the current state of design and implementation of fall detection devices. This review also examines the extent to which these devices have been tested in the real world as well as the acceptability of these devices to older adults.

Adults (aged > = 18 years of age)

Information not available

N = 57 (wearable systems)

Most common types of devices:

• Systems with device on trunk. Median sensitivity = 97.5% (range 81–100). Median specificity = 96.9% (range 77–100)

• Systems involving multiple sensors. Median sensitivity = 93.4% (range 92.5–94.2) and a median specificity of 99.8% (range 99.3–100).

• Systems involving devices around arms, hands, ears, or feet had a lower median sensitivity and specificity [81.5% (range 70.4–100) and 83% (range 80–95.7) respectively].

Silva de Lima et al 2017 [15]

N = 4 (2005–2015)

PubMed

Web of Science databases

To provide an overview of the use of wearable systems to assess freezing of gait (FOG) and falls in Parkinson’s disease with emphasis on device setup and results from validation procedures.

Parkinson disease patients (aged > = 18 years of age)

Average per study = 44 participants

Total = 177 participants

N = 2 (accelerometer)

N = 1 accelerometer and gyroscope)

N = 1 (accelerometer and force sensor)

High specificity (86.4–98.6%) and sensitivity (93.1% only one study) for wearable device detection of falls.

Rucco et al 2018 [11]

N = 42 (2002–2017)

IEEE Xplore

SpringerLink

Science Direct

PubMed

To provide an overview of the most adopted sensing technologies in these fields, with a focus on the type of sensors (rather than algorithms), their position on the body and the kind of tasks they are used in.

Healthy “aged” population

Average per study = 32 participants

Total = 1331 participants

N = 12 (accelerometer)

N = 7 (accelerometer and gyroscope)

N = 6 (accelerometer and pressure sensors)

N = 3 (accelerometer + another device)

N = 1 (gyroscope)

N = 4 (camera or radar or console)

N = 9 (three or more devices)

• Single sensor = 70% use accelerometer

• Two sensors = 1) Approaches that combine accelerometer with a pressure sensor (usually in shoes). 2) Approaches that use accelerometer and gyroscope sensors (usually on same electronic board).

• Three or more sensors = other sensing technology used (magnetometer, camera, EMG).

• Sensor placement = mainly on the trunk. Second most likely position is foot or leg (about 30%).

Sun et al 2018 [14]

N = 22 (2011–2017)

PubMed

Web of Science

Cochrane Library

CINAHL

To systematically evaluate the use of technology in performing fall risk assessments, and more specifically, to evaluate the test, sensor, and algorithm effectiveness on predicting and/or discriminating older adult fallers from non-fallers.

Older adults (Aged > 60 years of age)

Average per study = 86 participants

Total = 1896 participants

N = 11 (accelerometer)

N = 4 (accelerometer and gyroscope)

N = 4 (console)

N = 1 (laser)

N = 2 (accelerometer and pressure sensor)

A diverse range of diagnostic performance was observed (Accuracy: 47.9–100%, Sensitivity: 16.7–100%, Specificity: 40–100%, AUC 0.65–0.89) for wearable device detection of falls.