This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Aging, is properly cited. The complete bibliographic information, a link to the original publication on https://aging.jmir.org, as well as this copyright and license information must be included.
Falls and the risk of falling in older people pose a high risk for losing independence. As the risk of falling progresses over time, it is often not adequately diagnosed due to the long intervals between contacts with health care professionals. This leads to the risk of falling being not properly detected until the first fall. App-based software able to screen fall risks of older adults and to monitor the progress and presence of fall risk factors could detect a developing fall risk at an early stage prior to the first fall. As smartphones become more common in the elderly population, this approach is easily available and feasible.
The aim of the study is to evaluate the app Lindera Mobility Analysis (LIN). The reference standards determined the risk of falling and validated functional assessments of mobility.
The LIN app was utilized in home- and community-dwelling older adults aged 65 years or more. The Berg Balance Scale (BBS), the Tinetti Test (TIN), and the Timed Up & Go Test (TUG) were used as reference standards. In addition to descriptive statistics, data correlation and the comparison of the mean difference of analog measures (reference standards) and digital measures were tested. Spearman rank correlation analysis was performed and Bland-Altman (B-A) plots drawn.
Data of 42 participants could be obtained (n=25, 59.5%, women). There was a significant correlation between the LIN app and the BBS (
The digital app LIN has the potential to detect the risk of falling in older people. Further steps in establishing the validity of the LIN app should include its clinical applicability.
German Clinical Trials Register DRKS00025352; https://tinyurl.com/65awrd6a
As part of the aging process, older adults are affected by an increasing risk of falling as well as accidental falls [
As scientific evidence on the validity of such apps is limited, the aim of this explorative study was to evaluate the app LIN in comparison to established and validated functional assessments of mobility as a reference standard.
In 2021, this explorative validation study was conducted in Germany by the Geriatrics Research Group of Charité – Universitätsmedizin Berlin. The study was approved by the Ethics Committee of Charité – Universitätsmedizin Berlin (#EA1/363/20; date of approval: April 4, 2021). A sample size calculation was not performed as the study was exploratory in nature.
Participants were recruited from 3 sources: (1) the Geriatrics Research Group database, comprising older people who gave their consent to be contacted for participation in research projects; (2) older people who were staying in a geriatric hospital or day-care facility; and (3) a group of nursing home residents. Contact was made by mail, telephone, or a personal interview on-site. Inclusion criteria were age 65 years or older, being able to walk, and getting up from a chair and sitting down again. Participants were allowed to use walking aids, such as a wheeled walker or crutches. Exclusion criteria were defined as any fall events in the week before recruitment, more than 3 fall events during the past 6 months, and incapability of giving consent.
Data collection was conducted in the laboratory of the Geriatrics Research Group as well as in a nursing home and 2 day-care facilities. In addition to sociodemographic data, such as age and gender, the care level, degree of disability, data of mobility, and fall risk of the participants were recorded. The official care level within the German health care system ranges from level 0 (no need for care) to level 5 (maximum need for care)—§61b (1) German Social Code (SGB) XII, where SGB refers to the German Social Code. The official level of disability is characterized by level 20 (low disability) to level 100 (maximum disability)—§2 SGB IX. In addition, 4 mobility tests were performed, 3 reference assessments and LIN. In all measurements, LIN was used first. For this, participants filled out the app’s questionnaire independently or, if preferred, together with the researcher. A video of the patient’s gait was recorded using LIN on a smartphone. In a second step, 3 reference assessments were used to test the participants’ fall risk and mobility restrictions. Between assessments, the participants could rest by answering the questionnaire on sociodemographic data. All data were collected within 1 session.
LIN version 10.3.0 was used to determine the fall risk by computing a fall risk score. Input parameters to compute the fall risk score included (1) video analysis of each participant’s gait through an artificial intelligence–based algorithm [
The assessment was conducted with a mobile app using a smartphone with an integrated camera. The fall risk score is the weighted sum of 14 fall risk factors, as defined by the German National Expert Standard Fall Prevention [
The results of the gait analysis and the questionnaire were computed into a score of 0-100 points, with a higher scoring indicating a higher fall risk.
The technical validity of LIN has been described elsewhere in several publications [
The scientific approach underlying the app is based on a modular algorithm consisting of a video tester, a skeleton estimator (skeleton estimator 2D, skeleton estimator 3D, skeleton optimization 3D), and an analysis of mobility parameters. The skeleton estimator plays a central role. Both the validity of the mobility parameters and the validity of the analysis substantially depend on the spatial and temporal precision of the skeleton estimator [
Questionnaire: example of a person-related risk factor.
Clinical guidelines recommend the evaluation of gait or balance disturbances to detect fall risk, but there is no gold standard for assessing the risk of falling in older adults measuring functional abilities [
TUG is a short-duration simple test on mobility [
The BBS and TIN are scored based on a person’s ability to perform specific tasks. The BBS was developed in 1989 to determine balance stability among older adults [
A score below 45 points indicates a higher risk of falls [
TIN, also called Performance-Oriented Mobility Assessment (POMA), is a clinical balance assessment tool originally developed for use with institutionalized patients. It measures both balance and gait performance. Several versions of TIN are available, with varying numbers of items and score ranges [
Baseline and sociodemographic data were collected, and Spearman rank correlation analysis was conducted.
Additionally limits of agreement (LOA) between LIN and TIN, the BBS, and TUG were evaluated using Bland-Altman (B-A) plots [
As the Shapiro-Wilk tests revealed mostly nonnormal distributions for the calculated differences between the measurements, we used the median and defined the upper and lower 95% of the sorted results as the threshold instead of the ±1.96 SD used for B-A plots with normal-distributed data. This approach was recommended by Bland and Altman [
Baseline and sociodemographic data as well as all correlation analyses were calculated using SPSS Statistics version 28 (IBM Corporation, Armonk, NY, USA). All B-A plots were drawn using Microsoft Excel 2016.
Data of 42 participants, with a mean age of 77.6 (SD 7.3) years were analyzed. As can be seen in
One participant was not able to perform TUG due to difficulty in rising from the chair. Additionally, in 3 cases, data from LIN could not be interpreted and had to be discarded. Therefore, all correlation analyses were performed and B-A plots drawn with 39 and 38 data sets, respectively.
As can be seen in
In
Low scores for TUG indicated a high degree of functional mobility, while for the BBS and TIN, high scores indicated a high degree of mobility, and low scores for LIN indicated a low level of fall risk.
The results of LIN demonstrated a high correlation with the BBS (
As can be seen in
Baseline data.
Characteristics | Participants | |
Age (years; N=42), mean (SD) | 77.6 (7.3) | |
Female gender (N=42), n (%) | 25 (59.5) | |
|
||
|
No level | 26 (65) |
|
<30 | 1 (2.5) |
|
31-60 | 8 (20.0) |
|
61-80 | 5 (12.5) |
|
>80 | 0 |
|
||
|
0 | 25 (59.5) |
|
1 | 2 (4.7) |
|
2 | 7 (16.7) |
|
3 | 7 (16.7) |
|
4 | 1 (2.4) |
|
5 | 0 |
aThe official level of disability is characterized by level 20 (low disability) up to level 100 (maximum disability)—§2 German Social Code (SGB) IX.
Mobility data.
Assessment | Mean (SD) | Minimum | Maximum |
TUGa (N=40) | 13.7 (5.8) | 6.9 | 36 |
TINb (N=42) | 23.9 (5.3) | 8 | 28 |
BBSc (N=42) | 44.7 (13.0) | 7 | 56 |
LINd (N=39) | 19.8 (12.4) | 5 | 68 |
aTUG: Timed Up & Go Test.
bTIN: Tinetti Test.
cBBS: Berg Balance Scale.
dLIN: Lindera Mobility Analysis.
Spearman rank correlation of analog and digital fall risk and mobility assessment.
Assessment | TUGa | TINb | LINc | |||||||
|
|
N |
|
N |
|
N | ||||
BBSd | –0.770e | .001 | 40 | .730e | .001 | 42 | –0.611e | .001 | 39 | |
TUG | N/Af | N/A | N/A | –0.526e | .001 | 40 | .583e | .001 | 38 | |
TIN | N/A | N/A | N/A | N/A | N/A | N/A | –0.563e | .001 | 39 |
aTUG: Timed Up & Go Test.
bTIN: Tinetti Test.
cLIN: Lindera Mobility Analysis.
dBBS: Berg Balance Scale.
eThe correlation was significant at the level of .01.
fN/A: not applicable.
B-A plot of LIN and the BBS. B-A: Bland-Altman; BBS: Berg Balance Scale; LIN: Lindera Mobility Analysis.
B-A plot of TIN and LIN. B-A: Bland-Altman; LIN: Lindera Mobility Analysis; TIN: Tinetti Test.
B-A plot of TUG and LIN. B-A: Bland-Altman; LIN: Lindera Mobility Analysis; TUG: Timed Up & Go Test.
The aim of this study was to evaluate the accuracy of LIN compared to reference standards for analog objective measures of older people’s fall risk. As our study shows, a moderate-to-high correlation according to Cohen [
The results of our correlation analyses were verified by the B-A plots drawn. The B-A plots showed only a minority of the data pairs outside the predefined 95% limits. However, we observed a low-to-moderate proportional bias of the differences between results of LIN and the respective reference standards, indicating that both respective measurements might not be depicting the same construct. Moreover, we observed a skew in all plots, validating the observation of the correlation analyses. Due to the range and direction of the scales indicating a higher fall risk, we needed to transform our data for 2 plots in order to be able to obtain interpretable results. Additionally, as the differences between measurements were not normally distributed, we had to draw our B-A plots based on a nonparametric version.
This might have contributed to the results of the drawn plots. However, results from both correlation analyses and B-A plots could be interpreted as a sign that LIN can actually be superior in detecting older people at risk of falling compared to the 3 reference standards.
All 3 assessments are established tools for predicting falls in older people; however, none of them can be labeled as a gold standard. Although there might be different reasons for this, all of them have known flaws that have to be considered when planning to use any of them. As mentioned before, there are several versions available for TIN, making comparison between studies difficult. Additionally, both TIN and the BBS demonstrate only good but not high sensitivity and specificity for fall prediction in older adults living in care residence facilities [
This merits some consideration. In contrast to TUG, LIN, TIN, and the BBS record complex movement sequences and thus evaluate balance, postural control, and gait symmetry.
In contrast, TUG merges all these functional requirements into 1 single information piece, the time needed to complete TUG. As a consequence, a lot of technology-based research aims at increasing the information value gathered through the relative easy-to-administer TUG, where TUG performance is often used to gather not only the TUG time but also the TUG stride length, as well as the forward und lateral tilt of the trunk and gait symmetry. Although TUG’s ability to predict falls in older adults has been established [
Additionally, LIN uses an additional questionnaire based on the German National Expert Standard Fall Prevention and as such provides a guideline for the prevention of falls [
Despite these limitations, we deem our results satisfactory. The low number of data pairs outside the LOA indicate, in our estimation, a satisfactory level of comparability of the results of LIN with our reference standards. The observable bias in all 3 plots is, in our estimation, acceptable. Due to the reason stated before insofar, a complete agreement between the measurements cannot be expected. However, we are aware of the fact that the results presented here have to be interpreted with caution and have to be verified in further studies.
Compared to other apps for fall risk analysis, such as FallSA [
Using LIN or other medical devices with the ability to identify fall risks in older people while involving health professionals offers great potential. In 2021, Meekes et al [
Additionally, as several studies have demonstrated that a significant portion of patients tend to underestimate their own fall risk [
Using LIN has the potential to enable older people to be more independent of the initial determination of a fall risk by GPs or other health care professionals and also enables them to identify and respond to positive or negative changes in their own fall risk. This provides older adults with the ability to manage their own fall risk in an effective and adequate manner. Using LIN can help reduce fall events in people aged 65 years or more. Further study is indicated to verify validity.
Bland-Altman
Berg Balance Scale
general practitioner
Lindera Mobility Analysis
limits of agreement
German Social Code
Tinetti Test
Timed Up & Go Test
We thank Lindera GmbH for providing the app Lindera Mobility Analysis (LIN) for the study. We acknowledge support from the German health insurance company Allgemeine Ortskrankenkasse (AOK).
We also acknowledge financial support from the Open Access Publication Fund of Charité – Universitätsmedizin Berlin and the German Research Foundation (DFG).
The authors report that there are no competing interests to declare. The study was commissioned by Lindera GmbH. Lindera GmbH did not influence the results presented in this study.