Remember me
Background:
Neurodegenerative diseases (NDs) are a significant threat to human health. Numerous research demonstrated that patients with NDs might present with decreased balance, which is responsible for an increased risk of falling. As an emerging technology, wearable devices can detect falls and prevent privacy breaches.
Objective:
To access the evolution of trends and technology in wearable devices to detect falls among patients with NDs.
Methods:
We screened PubMed and Web of Science (February 2023) to summarize the pathway of fall detection with any body-worn sensor. Included articles were required to be full-text and published in English. Documents were excluded if they; (1) only used wearable devices for fall cueing, (2) did not offer sufficient information for data extraction, (3) did not use patients with NDs, (4) only used non-wearable sensors or devices.
Results:
The review identified 89 articles at the end of the procedure for data extraction. A wide variety existed in participant sample size (1–131), sensor types, placement and algorithms. 97.75% of papers (n = 87) used patients with Parkinson’s disease as experimental subjects. 21.45% of studies attached devices on the ankle (n = 19), with a clear preference for using multiple types of sensors (58.43% of studies, n = 52). As the most commonly used inertial measurement unit (IMU), 21 articles utilized accelerometers and gyroscopes to assess falls. 39.33% of studies (n = 35) choose data set to verify the effectiveness of their algorithm. Machine learning algorithms have become prevalent since 2019, and the most commonly used algorithm was support vector machine (SVM) (n = 17).
Conclusion:
These results show that an increasing number of researchers examine the validation performance of their systems in non-real-time. The ankle was the preferred location among researchers, and there is a clear preference to use multiple types of sensors and machine learning algorithms to improve accuracy and immediacy. Future work should focus on other NDs instead of limiting to Parkinson’s disease and consider an adequately studied population. A consensus on walking tasks and accuracy measurements is urgently needed. Performing studies in a simulated free-living environment for a specified time frame is advisable, with continuous real-time monitoring and assessment.
Systematic review registration:
PROSPERO, identifier (CRD42023405952).
IntroductionAs a major threat to human health (Dommershuijsen et al., 2020; Kingwell, 2019), NDs (e.g., Parkinson’s disease, Alzheimer’s disease, motor neuron disease, and dementia) comprise a heterogeneous group of neurological conditions characterized by progressive—and often currently incurable—clinical courses. With the ongoing extension of lifespan, the prevalence and societal burden of these age-dependent disorders continue to rise (Heemels, 2016). Patients with NDs commonly exhibit motor impairments as well as cognitive and behavioral disturbances (Pender et al., 2020; Aarsland et al., 2021), which may manifest as impaired postural control, gait abnormalities (Morel et al., 2020), and consequently an elevated risk of falls (Schell et al., 2019). Falls in this population are not only associated with fractures, hospitalization, and loss of independence, but may also precipitate secondary complications (e.g., fear of falling and reduced mobility), thereby accelerating functional decline. Therefore, developing accurate and practical fall-detection solutions is of clinical importance to reduce injury-related morbidity and the downstream costs of post-fall care.
Wearable sensing has emerged as a promising approach for fall detection because sensors can be worn continuously and capture movement signals in everyday contexts when deployed at appropriate body locations. Compared with many environmental approaches, wearable solutions can support monitoring across both indoor and outdoor settings while offering a more privacy-preserving pathway for continuous assessment. Nevertheless, key barriers remain, including limited battery life, susceptibility to false alarms, and user adherence—factors that directly condition real-world feasibility even when laboratory performance is acceptable. Recent advances in mobile and embedded technologies have enabled miniaturized, energy-efficient devices with improved on-device processing and wireless connectivity, which can facilitate timely alerts and potentially mitigate adverse outcomes related to prolonged “long-lie” after a fall. Moreover, wearable platforms may function as personalized monitoring tools by providing quantitative, longitudinal information relevant to disease severity and mobility impairment, while reducing reliance on intrusive sensing modalities.
Despite substantial growth in the literature, fall-detection research in NDs populations remains methodologically heterogeneous, spanning diverse sensor modalities, placements, algorithms, and validation protocols, which contributes to fragmented evidence and limited cross-study comparability. Recent reviews in Ambient Assisted Living and Human Activity Recognition and wearable assisted-living have summarized broader wearable fall-detection advances and highlighted practical design constraints (e.g., unobtrusiveness, miniaturization, energy efficiency, and privacy) (Guerra et al., 2023; Li et al., 2025; Iadarola et al., 2024). Performance-oriented syntheses further underscore that validation performance is central to viability (Gorce and Jacquier-Bret, 2025). However, a NDs-focused synthesis that explicitly tracks how wearable fall detection has evolved over time—and that systematically compares validation performance across heterogeneous technological and methodological choices—remains limited. Accordingly, we conducted a systematic review to examine the temporal evolution of wearable-sensor fall detection in NDs populations in terms of sensor technology, body placement, algorithmic strategies, and validation performance, with the aim of clarifying robust evidence, improving comparability, and informing priorities for future investigations.
Review methodologyA systematic literature review was conducted in light of the PRISMA statement (Liberati et al., 2009). We searched PubMed and Web of Science in February 2023 to summarize fall detection using body-worn sensors in patients with NDs. These databases were selected to allow both engineering and medical journals to be included during the search procedure. Additionally, a search in the reference of review articles and book chapters that appeared during the search was performed. The objective was to identify potentially eligible studies absent in the database search. The final search query is summarized in Table 1.
DatabaseSearch stringRecordsWeb of Science#1:(((((((((((((((((((((TS = (Parkinson*)) OR TS = (PD)) OR TS = (Paralysis Agitans)) OR TS = (Alzheimer)) OR TS = (ATD)) OR TS = (Dementia, Senile)) OR TS = (Senile Dementia)) OR TS = (Primary Senile Degenerative Dementia)) OR TS = (Dementia, Primary Senile Degenerative)) OR TS = (Dementia, Presenile)) OR TS = (Presenile Dementia)) OR TS = (Sclerosis, Amyotrophic Lateral)) OR TS = (ALS)) OR TS = (Gehrig’s Disease)) OR TS = (Gehrig Disease)) OR TS = (Gehrigs Disease)) OR TS = (Charcot Disease)) OR TS = (Guam Disease)) OR TS = (Disease, Guam)) OR TS = (motor neuron diseases)) OR TS = (Lou-Gehrigs Disease)) OR TS = (Disease, Lou-Gehrigs)1,852#2: (TS = (fall*)) OR TI = (fall*)#3: (((((TS = (sensor*)) OR TI = [13]) OR TS = (wearable*)) OR TI = (wearable*)) OR TS = (device*)) OR TI = (device*)#1 AND #2 AND #3PubMed#1: “parkinson*”[Title/Abstract] OR “PD”[Title/Abstract] OR “paralysis agitans”[Title/Abstract] OR “alzheimer*”[Title/Abstract] OR “ATD”[Title/Abstract] OR “dementia senile”[Title/Abstract] OR “senile dementia”[Title/Abstract] OR “primary senile degenerative dementia”[Title/Abstract] OR “dementia primary senile degenerative”[Title/Abstract] OR “dementia presenile”[Title/Abstract] OR “presenile dementia”[Title/Abstract] OR “amyotrophic lateral sclerosis”[Title/Abstract] OR “sclerosis amyotrophic lateral”[Title/Abstract] OR “ALS”[Title/Abstract] OR “motor neuron diseases”[Title/Abstract] OR “gehrig s disease”[Title/Abstract] OR “gehrig disease”[Title/Abstract] OR “charcot disease”[Title/Abstract] OR “guam disease”[Title/Abstract] OR “disease guam”[Title/Abstract]484#2: “fall*”[Title/Abstract]#3 “wearable*”[Title/Abstract] OR “sensor*”[Title/Abstract] OR “device”[Title/Abstract]#1 AND #2 AND #3Search string used for each database.
The truncation symbol was used to broaden the search with more specificity.
We included articles if they were full-text, published in English, and published in a peer-reviewed journal. In the meantime, involved papers should focus on fall detection or fall-risk assessment using wearable (body-worn) sensors in NDs populations, and present original research validating wearable sensors to assess falls or fall risk. We excluded articles if they; (1) only used wearable devices for fall cueing, (2) did not offer sufficient information for data extraction, (3) did not use patients with NDs, or (4) only used non-wearable sensors or devices. Algorithm performance metrics were not used as eligibility criteria; when reported, they were extracted and synthesized in the Results.
YC and TH finalized the standard of inclusion and exclusion, then independently screened the title, abstract and keyword in the databases. Repetitive outcomes were filtered out, and the remaining articles were relevant following their title and abstract. The remaining papers were reviewed in full document, and the applicable data was extracted from identified studies and tabularized under the pre-established heading. Divergences between reviewers were resolved by consensus. For each included study, we extracted the following variables: author(s), studied population, sensor type, device location(s) (including the number of placements, n), walking task, method category (threshold-based, machine learning, or deep learning), specific classifier/model, reported performance metrics, evaluation mode (online [ON] vs. offline [OFF]), publication year, real-time implementation (yes/no), and data source (e.g., public dataset vs. self-collected data).
ResultsStudies selectionThe electronic database searches yielded 2,336 results that fulfilled the requirements for inclusion (Figure 1). Simultaneously, surveying the literature cited in these papers allowed for the identification of 10 more documents were included. Four hundred forty-six manuscripts were dismissed as duplicates, leaving 1890 papers being screened (1,635 records excluded). Of the remaining of 255 articles were filtered by full document. Eighty-nine articles were deemed relevant for this review.
Study flow diagram.
This review analyzed the application of wearable sensors to access falls in patients with NDs (Table 2). Of all 89 articles, 87 articles used patients with Parkinson’s disease (PD), while seven and two articles recruited healthy elderly control and healthy control, and only one study enrolled neurological disorders sufferers. Concurrently, the enrollment count of fall detection projects ranged in complexity (range, 1–131, median = 14). Nevertheless, 39.33% of articles (n = 35) leverage data sets to appraise their algorithms’ credibility (Figure 2). Data from Bachlin et al. (2010) was the most frequently used data set (45.71% of studies, n = 16).
AuthorStudied populationType of sensorDevice location (n)Walking taskMethod categoryClassifier/modelPerformance (reported metrics)ONOFFYearReal timeSource of datasetAhlrichs et al. (2016)20 PDAccelerometer; Gyroscope; MagnetometerWaist (1)Scripted activities simulating natural behavior at the patients’ homeMachine learningSVMSensitivity: 92.3%; Specificity: 100.0%––2016YMartín et al. (2015)Ahn et al. (2017)10 PDAccelerometer; Gyroscope; MagnetometerHead (1)
Comments (0)