Screening of apathetic elderly adults using kinematic information in gait and sit-to-stand/stand-to-sit movements measured with Doppler radar

Abstract

This paper presents a Doppler radar apathy-screening technique applied to elderly people based on their basic daily activities of walking and movements of sit-to-stand and stand-to-sit (STS). Our Doppler radar system remotely measured the kinematic parameters of the movements of 78 community-dwelling elderly adults (27 apathetic participants and 51 non-apathetic ones). Subsequently, logistic regression models using the measured kinematic parameters of gait and sit-to-stand/stand-to-sit movements were constructed for screening. The experimental results verified that, although the model using gait parameters could screen an apathetic group with a sensitivity of 85.2% and a specificity of 58.8%, the model using the STS parameters achieved better screening accuracies with a sensitivity of 88.9% and a specificity of 76.5%. These results reveal that the kinematic information of STS movements is significantly more effective at detecting apathy than is the gait information, which is otherwise regarded to be effective in conventional epidemiological studies.

Keywords

apathy screening Cybercare Doppler radar gait analysis motion analysis

Introduction

Apathy, which is a lack of motivation and interest in life activities,^1,2 is a major symptom of cognitive impairment and depression. Thus, early detection is important for the prevention and caring for their neuropsychological disorders (e.g., Parkinson’s disease, schizophrenia, and Alzheimer’s disease). To diagnose apathy, a subjective assessment by a doctor based on the observation of facial expressions, patient attitudes, and a paper-based questionnaire (i.e., the Apathy Scale) is generally used.^3,4 However, such assessments are not suitable for daily monitoring or early detection because questionnaires of the Apathy Scale are fixed, and a grader is required. Thus, the daily objective evaluation of apathy would be difficult, even in clinical environments.

For the early detection of apathy via unobtrusive daily monitoring methods, sensor-based techniques have been examined.⁵ Some studies have reported that actigraphy is an effective approach.^5,6 However, the subjects must wear sometimes bulky sensors. Speech analysis⁷ and face-image processes⁸ can achieve automatic and remote screening. However, these techniques involve privacy issues and the performance of specific tasks that often constrain the subjects. To resolve these problems, gait-based apathy detection^5,9 has been shown as a promising approach, because significant associations between gait speed and apathy have been found in epidemiological studies.^7–9 To measure gait speed, infrared⁵ and pressure sensors⁹ have been used, although their sensing areas are limited. Moreover, conventional studies have clarified only the associations between apathy and gait speed, and a concrete method to screen apathetic people has not been considered. Furthermore, most studies have considered only gait speed without considering other gait parameters (e.g., leg velocities) and related motions during daily activities (e.g., standing-up and sitting-down movements). For the measurement of motion parameters in daily activities, wearable sensors, such as accelerometers, have also been studied. However, they require subjects to wear sometimes burdensome devices.¹⁰ Optical cameras and depth sensors are other possible methods,^11,12 owing to their remote-sensing capabilities. However, they are all affected by light conditions and clothing materials. Furthermore, these sensor techniques cannot directly obtain velocity parameters for the acquisition of detailed motion information.¹³

To screen for apathy based on accurate remote velocity measurement without the abovementioned restrictions of the accelerometers and optical sensors, micro-Doppler radar technologies have been used to measure walking and sit-to-stand/stand-to-sit (STS) movements. Notably, their application to the screening of apathetic elderly adults is promising. The Doppler radar technique achieves remote measurement of motion information using a small number of fixed sensors.^14–16 Some previous studies have shown that the detailed gait parameters measured using Doppler radar techniques were effective in assessing cognitive impairment,¹⁵ and the kinematic parameters of STS movements could be used to screen impairments based on a participant’s ability to perform instrumental activities of daily living (IADL) in modern settings and with modern tools (e.g., smartphones).¹⁶ Our research hypothesis is that apathy could be screened using the Doppler radar measurements of gait and STS movements based on significant findings of correlation between apathy and cognitive decline¹⁷ or IADL impairment.¹⁸

This paper demonstrates the capability of screening apathy in elderly participants using Doppler radar measurements of gait and STS movements. The participants were 75 years and older, because the prevalence of frailty is significantly increased for individuals aged 75 years and older.¹⁹ This frailty is closely related to apathy.²⁰ We initially extracted the kinematic parameters of gait and STS movements and investigated their differences between nonapathetic and apathetic participants. Furthermore, to investigate the effectiveness of apathy screening using combinations of extracted parameters, two logistical regression models were separately constructed using the gait and kinematic parameters of STS movements. The screening capability of each extracted parameter and the constructed models was evaluated and compared using receiver operating characteristic (ROC)-curve analyses. The results showed that, although the gait-based screening model achieved apathy screening to some extent, the model constructed using the STS parameters achieved more precise accuracy. Thus, the significant associations between apathy and the kinematic information of STS movements measured using the Doppler radar are revealed. These finding can enable the development of unconstrained screening systems for the apathetic elderly.

Methods

Study participants and experimental procedure

The study participants comprised of 78 community-dwelling adults aged 75 years and older (38 men and 40 women, age 78.6 ± 3.71 years). All participants could perform walking and STS movements without the assistance of other people or walking aids. None of the participants had been diagnosed with cognitive impairment or dementia.

The participants first completed the Apathy Scale, a paper-based test.³ Next, radar measurements of participants’ walking¹⁵ and STS movements¹⁶ were taken to extract kinematic parameters. The extracted parameters, after correlation to the Apathy Scale results, were used to screen for apathy. The participants completed other paper-based tests to evaluate cognitive functions and IADLs for additional background and screening. The cognitive function test was the mini-mental state examination (MMSE),²¹ and the IADL test was the Japan Science and Technology Agency Index of Competence (JST-IC).²² Participants were provided with written and verbal instructions regarding the testing procedures, and written consent was obtained from each prior to testing.

Apathy scale

The Apathy Scale was first used to determine apathetic participants. It is a paper-based test containing 14 questions. Examples of questions include “are you interested in learning new things?” “are you concerned about your condition?” and “do you have the energy for daily activities?” All questions are listed in Ref. 2. The participants responded to each by selecting one of four levels of answer: “a lot,” “some,” “slightly,” and “not at all.” An evaluation score between zero and three was given for each response. The summation of all questions became the total score (maximum 42). A higher score (16+) indicates a higher apathetic condition.³

Micro-Doppler radar measurements

This study used kinematic parameters of gait and STS movements extracted by means of Doppler radar measurements. The radar systems were similar to those used in previous studies, including the gait-measurement system of Saho et al.¹⁵ and the STS measurement system of Saho et al.¹⁶ Figure 1 shows the experimental site for both measurement systems. For both, the radars transmitted a 24-GHz sinusoidal wave with an effective isotropic radiated power of 40 mW. The directivity in the horizontal plane was ±35°, and that in the vertical plane was ±14°. We sat up these experimental systems in one room and performed experiments with each participant. Only one participant was in the beam illumination area at a time. Signals were received by demodulating the reflected waves with the sampling frequency of the received signal being 600 Hz. The received signals comprised the Doppler frequencies corresponding to the velocities of the scatters on each body part, and these were expressed as a spectrogram using a short-time Fourier transform (STFT) method. Examples of the spectrograms are shown in Figure 2.

Figure 1.

Radar experimental site: (a) gait measurement and (b) measurement of STS movements.

Figure 2.

Examples of the spectrograms and outlines of feature parameter extraction: (a) gait and (b) STS movements.

In the gait measurement of Figure 1(a), the radar was placed just in front of the walking participant at a height of 0.86 m. The participants walked toward the radar along a 10-m walkway at a self-selected, comfortable pace. Figure 2(a) shows an example of the STFT spectrogram of the walking test. From the spectrograms, kinematic information was extracted as motion parameters. In a similar fashion to Saho et al.,¹⁵ we extracted the mean, upper, and lower envelopes, V_m(t), V_u(t), and V_l(t), respectively, as indicated in Figure 2(a), and calculated six parameters using their means and standard deviations: v_m,mean = E[V_m(t)], v_m,std = STD[V_m(t)], v_u,mean = E[V_u(t)], v_u,std = STD[V_u(t)], v_l,mean = E[V_l(t)], and v_l,std = STD[V_l(t)]. E[ ] and STD [ ] indicate means and standard deviations with respect to time t. For example, we can see the cyclic variation of velocities corresponding to body motion in V_m(t) from Figure 2(a). v_m,mean and v_m,std thus correspond to walking speed and degree of variation of body velocities, respectively. Table 1 shows the list of these parameters.

Table 1.

List of extracted kinematic parameters.

Movement	Parameter	Meaning
Walking	v _m,mean	Mean walking speed
	v _m,std	Degree of variation of body velocities
	v _u,mean	Mean speed of leg forward motion
	v _u,std	Degree of variation of velocities of legs in leg forward motion
	v _l,mean	Mean speed of leg motion in stance phase
	v _l,std	Degree of variation of velocities of legs in stance phase
Sit-to-stand	T _sta	Duration from sitting to standing-up
	v _sta,mean	Mean velocity from sitting to standing-up
	v _sta,max	Maximum velocity from sitting to standing-up
	a _sta,mean	Mean acceleration from sitting to standing-up
	a _sta,max	Maximum acceleration from sitting to standing-up
	a _sta,min	Minimum acceleration from sitting to standing-up
Stand-to-sit	T _sit	Duration from standing to sitting-down
	v _sit,mean	Mean velocity from standing to sitting-down
	v _sit,min	Minimum velocity from standing to sitting-down
	a _sit,mean	Mean acceleration from standing to sitting-down
	a _sit,max	Maximum acceleration from standing to sitting-down
	a _sit,min	Minimum acceleration from standing to sitting-down

In the measurement of the STS movements of Figure 1(b), the radar was placed just above the participants as they were standing. The height of the radar was 3 m, and the height of the seat was 0.43 m. The participants first sat down on a chair and were instructed to stand up and sit down again at a comfortable speed. Figure 2(b) shows an example of the spectrogram of the STS movements. We extracted the median velocity components, V_c(t), and calculated 12 kinematic parameters using V_c(t) with the same procedure specifically described by Saho et al.¹⁶ The list of the extracted parameters is shown in Table 1.

Apathy screening model construction

We first considered apathy screening using the cutoff value of each parameter. To specify those that indicated significant differences between the participants in apathy and not apathy groups, we applied Welch’s t-test using a significance level of p = 0.05. For parameters having p < 0.05, we performed ROC-curve analyses, and their screening capabilities were evaluated using the area under the ROC curve (AUC).²³ In the ROC-curve analysis, relationships between the false- (1-specificity) and true-positive rates (sensitivity) were calculated by changing the cut-off value in the screening of apathy using each selected parameter. This relationship is the ROC curve. For example, if the difference between the apathetic and nonapathetic groups of v_u,mean was significant, we set a cut-off value, v_u,mean,cut, and classified the participants with v_u,mean < v_u,mean,cut as the apathetic group. Then, the false- and true-positive rates for all v_u,mean,cut, for which different rates were obtained, were calculated to depict the ROC curve. We defined the screening capability of the selected parameter as the AUC calculated by the integral of the ROC curve. The AUC is equal to the probability that the selected parameter will rank a randomly chosen apathetic participant with a lower apathetic tendency than one given to a randomly chosen nonapathetic participant. Thus, the AUC indicates the screening capability of the selected parameters, considering all possible cut-off values. The maximum value of the AUC is one, and a higher AUC indicates the good screening capability of the selected parameter.

We then constructed logistic regression models to screen the participants using the parameters shown in Table 1 and compared their AUCs with the abovementioned models constructed using each parameter. The constructed model is expressed as p_ap = 1/{1+exp(-β₀-β₁x₁-β₁x₁-. . .)}, where p_ap is the probability that the participant is in the apathetic group, β_i is i-th coefficient, and x_i is i-th explanatory variable selected from the extracted parameters. We constructed and compared two models. p_ap,gait was constructed using the gait parameters based on the conventional finding of epidemiological studies of the significant relationship between apathy and gait, and p_ap,sts was constructed using the parameters of the STS movements to reveal the relationship between the STS movements and apathy. The selection of explanatory variables for both models used the lowest Akaike information criterion²⁴ from all combinations of the parameters, and the coefficients (β_i) were calculated via the maximum likelihood estimation technique.²⁴

The participants having apathy were screened using a cut-off value for p_ap,gait or p_ap,sts. Like the evaluations of screening capability for each parameter explained above, ROC-curve analyses were used for the constructed models for the same reason. For the constructed models of p_ap,gait and p_ap,sts, we considered the appropriate cut-off points for the screening using Youden’s index.²⁵ to clarify their sensitivity and specificity. The significance of the differences of AUCs was tested using the p-value of a DeLong’s test with a significance level of p = 0.05.²³

Results

Based on the Apathy Scale score, 27 participants were classified into the apathetic group (16 men and 11 women: age 77.9 ± 2.76 years, height 157.5 ± 8.80 cm, mass 56.9 ± 10.3 kg, MMSE score 26.1 ± 3.00, and JST-IC score 8.81 ± 3.39). 51 participants were classified into the nonapathetic group (22 men and 29 women: age 79.0 ± 4.09 years, height 154.8 ± 9.16 cm, mass 56.9 ± 10.3 kg, MMSE score 27.4 ± 2.28, and JST-IC score 11.8 ± 2.94). No significant differences in age, height, mass, or MMSE were found between the two groups. Thus, although the age,²⁶ height,²⁷ mass,²⁷ and cognitive status (MMSE score)²⁸ are related to the walking, sit-to-stand, and stand-to-sit motions, our results were not affected by them. The JST-IC score of the nonapathetic group was significantly larger, however, than that of the apathetic group (p < 0.001 with Welch’s t-test)

Table 2 summarizes the extraction results for all parameters, including the results of the t-tests and AUC evaluations. Significant differences between the two groups were found for certain parameters. However, the AUCs for the screenings using these parameters were not high (<0.7). Thus, accurate apathy screening using single kinematic parameters of the gait and STSs was difficult, and the appropriate cut-off values for these parameters were not determined. Thus, a combination of extracted parameters were considered to achieve accurate screening.

Table 2.

Summary of parameter extraction results and results for the t-tests and AUC calculations.

Parameters	Nonapathetic	Apathetic	p	AUC (for p < 0.05)
v_m,mean [m/s]	1.326 ± 0.189	1.252 ± 0.237	0.165
v_u,mean [m/s]	2.325 ± 0.260	2.178 ± 0.310	0.0366*	0.636
v_l,mean [m/s]	0.546 ± 0.0972	0.530 ± 0.0820	0.427
v_u,std [m/s]	0.113 ± 0.0216	0.109 ± 0.0235	0.556
v_l,mean [m/s]	0.364 ± 0.167	0.354 ± 0.158	0.803
v_l,std [m/s]	0.150 ± 0.0328	0.155 ± 0.0519	0.661
T_sta [s]	0.652 ± 0.150	0.724 ± 0.150	0.0469*	0.643
v_sta,mean [m/s]	0.470 ± 0.0904	0.467 ± 0.0918	0.865
v_sta,max [m/s]	0.797 ± 0.169	0.782 ± 0.170	0.706
a_sta,mean [m/s²]	0.0279 ± 0.0615	–0.00751 ± 0.0339	0.00159*	0.678
a_sta,max [m/s²]	4.929 ± 1.134	4.681 ± 1.258	0.396
a_sta,min [m/s²]	–3.134 ± 0.947	–2.668 ± 0.915	0.0390*	0.645
T_sit [s]	0.728 ± 0.183	0.814 ± 0.189	0.0496*	0.671
v_sit,mean [m/s]	0.409 ± 0.104	0.428 ± 0.0981	0.414
v_sit,min [m/s]	0.699 ± 0.196	0.707 ± 0.171	0.849
a_sit,mean [m/s²]	–0.0272 ± 0.0702	–0.00912 ± 0.0435	0.167
a_sit,max [m/s²]	–2.718 ± 1.118	–2.379 ± 0.757	0.117
a_sit,min [m/s²]	3.922 ± 1.011	4.087 ± 1.223	0.552

p < 0.05.

The model construction result using the logistic regression for p_ap,gait was (x₁, x₂) = (v_u,mean, v_l,std) and (β₀, β₁, β₂) = (4.13, −3.56, 21.4). That for p_ap,sts was (x₁, x₂, x₃, x₄, x₅, x₆) = (a_sta,mean, a_sta,max, T_sit, v_sit,mean, a_sit,max, a_sit,min) and (β₀, β₁, β₂, β₃, β₄, β₅, β₆) = (−6.48, −27.5, −1.03, 5.12, 22.6, 3.07, 1.29). For all coefficients, p-values in t-tests assuming the null hypothesis of β_i = 0 were smaller than 0.05. Figure 3 shows examples of the plots for the selected kinematic parameters. This figure shows that no clear boundaries between the two groups were found in either the gait or STS parameters. However, as indicated in Figure 3(b), we confirmed that the absolute values of the kinematic parameters of some participants in the non-apathetic group were large and could be used to predict the screening of apathetic adults using smaller values. Moreover, other selected parameters of the STS movements had similar trends.

Figure 3.

Extracted parameters for all participants: (a) gait parameters selected for p_ap,gait and (b) examples of selected parameters in STS movements.

Figure 4 shows the ROC curves with p_ap,gait and p_ap,sts. Their AUCs were 0.725 and 0.880, respectively, and the significant differences among these AUCs were verified with DeLong’s test, which resulted in p = 0.023. These AUCs were larger than that of Table 2, indicating the effectiveness of the constructed models based on the combination of extracted parameters. For p_ap,gait, the determined cut-off value was 0.290 with 85.2% sensitivity and 58.8% specificity (Youden’s index = 0.440). For p_ap,sts, the cut-off value was 0.367 with 88.9% sensitivity and 76.5% specificity (Youden’s index = 0.654).

Figure 4.

ROC curves calculated with the extracted kinematic parameters.

Discussion

Findings of this study

Our results demonstrate the successful screening capability of apathetic elderly adults using the Doppler radar measurement of basic daily activities (i.e., gait and STS movements). The findings of our study can enable the development of apathy-related risk-monitoring systems that operate ubiquitously without constraint. Based on the results of the t-tests and ROC analyses of the extracted parameters, several kinematic parameters were found to be significantly associated with apathy. The absolute values of the velocity and acceleration of the nonapathetic group were greater than those of the apathetic one, and the times required to perform the standing-up and sitting-down movements by the nonapathetic group were lesser than those of the apathetic group. These results reflect the deterioration of physical functions associated with the basic activities of daily living, which are known to be related to apathy. However, the AUCs for each parameter were not large, and accurate apathy screening using a single parameter was found to be difficult using the kinematic information of both gait and STS movements.

The screening capabilities of the logistic models constructed using combinations of the extracted parameters were investigated. The constructed p_ap,gait model exhibited screening capabilities to some extent. This was consistent with the results of epidemiological studies that investigated the associations between apathy and walking.^14,25 The contributions of our study from the viewpoint of gait-based apathy detection include a novel model for screening apathy behaviors based on gait information and a quantitative screening method.

An accurate screening of apathetic participants using the p_ap,sts model, compared with the gait-based model of p_ap,gait, was achieved. Table 2 presents the kinematic relationships between STS movements and apathy, including significant differences. These results are the first to reveal that the measurement of STS movements were more effective than gait measurement for apathy screening.

Mechanism of our results

Epidemiological studies have shown that the co-occurrence of apathy and gait can be caused by subcortical lesions.^9,29 The resultant neurological functioning deterioration leads to impairment of executive (cognitive) functions and planning ability related to motor control. This corresponds to apathy and is related to deteriorated gait speed and leg motion.^15,30 However, MMSE scores of both groups did not show significant differences. Our results indicated that the STS method achieved better screening accuracy. Motor-control functions are also related to the ability to perform STS movements,³¹ which leads to the association between measured kinematic parameters and reported apathy levels. A previous study revealed that IADL impairments could be detected using measurements of STS movements.¹⁶ Significant relationships between apathy and IADL impairment have since been reported.¹⁷ As indicated, the IADLs scores of the apathetic group were significantly smaller than those of the non-apathetic group. Thus, the linkage of apathy to IADL impairment might be accentuated, owing to the dysfunction of motor control related to apathy in the STS movements.

Furthermore, we found that acceleration was significant to apathy screening, because a relatively large number of acceleration parameters were selected for p_ap,sts. This can be explained by the apathy-related deterioration of the physical functions and the role of acceleration in STS movements. Significant relationships between apathy and physical functions (e.g., gait, balance, endurance, and body movements) have been reported.^32,33 These results indicate that physical functions deteriorate as apathy increases. It was, therefore, predicted that this deterioration would correspond to worsening STS performance. With respect to the elderly, some researchers have revealed that acceleration parameters are critical factors for the smoothness of STS movements, compared with velocity parameters.^34,35 Based on these reports, we can consider that our results indicated apathy-related deterioration of the physical functions of STS movements through their acceleration parameters. Thus, although the degree of change in STS movements related to apathy was not high and was not clearly reflected by the velocity parameters, the acceleration parameters that were more sensitive to even slight deteriorations in physical functions indicated the differences between the apathetic and nonapathetic groups in our results.

Conclusion

This paper investigated the kinematic information associated with physical impairment related to apathy and presented a new apathy-screening model constructed using kinematic parameters measured by Doppler radar systems. Based on the experimental results obtained from participants aged 75 years and older, we confirmed that gait parameters other than walking speed could be used for apathy screening and that significant associations existed between apathy and the kinematic parameters of STS movements. The model constructed using the STS parameters achieved accurate screening of the apathetic elderly adults with an AUC of 0.880 and was significantly better than the model constructed using gait parameters. These results make the development of an unconstrained daily apathy screening-system possible.

However, the number of participants was not large and was limited to people aged 75 years and older. This restricted our ability to apply advanced machine-learning approaches. To enhance the effectiveness of the presented methods and to resolve these problems, experiments involving larger numbers of participants of a wider range of ages (e.g., 65 years and older) and investigations based on machine-learning approaches should be undertaken in the near future. Moreover, our experiments were conducted for a situation having one radar experimental site that measured gait and STS movements (Figure 1). Evaluations of our method’s effectiveness in more realistic scenarios are important future tasks.

Footnotes

Acknowledgements

We would like to thank Editage () for English language editing.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethical approval

The experimental protocol was approved by the local ethics committee (Toyama Prefectural University, approval no. H29-1).

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported in part by the NAKAJIMA FOUNDATION.

ORCID iD

Kenshi Saho

References

Marin

RS.

Differential diagnosis and classification of apathy. Am J Psychiatry 1990; 147: 22–30.

Starkstein

Mayberg

Preziosi

, et al. Reliability, validity, and clinical correlates of apathy in Parkinson’s disease. J Neuropsychiatry Clin Neurosci 1992; 4: 134–139.

Okada

Kobayashi

Yamagata

, et al. Poststroke apathy and regional cerebral blood flow. Stroke 1997; 28: 2437–2441.

Radakovic

Abrahams

Developing a new apathy measurement scale: dimensional apathy scale. Psychiatry Res 2014; 219: 658–663.

König

Aalten

Verhey

, et al. A review of current information and communication technologies: can they be used to assess apathy. Int J Geriatric Psychiatry 2014; 29: 345–358.

Kluge

Kirschner

Hager

, et al. Combining actigraphy, ecological momentary assessment and neuroimaging to study apathy in patients with schizophrenia. Schizophrenia Res 2018; 195: 176–182.

König

Linz

Zeghari

, et al. Detecting apathy in older adults with cognitive disorders using automatic speech analysis. J Alzheimers Dis 2019; 69: 1183–1193.

Happy

Dantcheva

Das

, et al. Characterizing the state of apathy with facial expression and motion analysis. In: Proceedings of the 2019 14th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2019), Lille, France, 14–18 May 2019, pp.1–8. IEEE.

Ayers

Shapiro

Holtzer

, et al. Symptoms of apathy independently predict incident frailty and disability in community-dwelling older adults. J Clin Psychiatry 2017; 78: 529–536.

10.

Madigan

EA.

Fitness band accuracy in older community dwelling adults. Health Inform J 2019; 25(3): 676–682.

11.

Cronin

Doherty

Touchless computer interfaces in hospitals: a review. Health Informatics J 2019; 25(4): 1325–1342.

12.

Guffanti

Brunete

Hernando

Non-invasive multi-camera gait analysis system and its application to gender classification. IEEE Access 2020; 8: 95734–95746.

13.

Murmu

Chakraborty

Nandi

Relative velocity measurement using low cost single camera-based stereo vision system. Measurement 2019; 141: 1–11.

14.

Seifert

A-K

Amin

Zoubir

AM.

Toward unobtrusive in-home gait analysis based on radar micro-Doppler signatures. IEEE Trans Biomed Eng 2019; 66: 2629–2640.

15.

Saho

Uemura

Sugano

, et al. Using micro-Doppler radar to measure gait features associated with cognitive functions in elderly adults. IEEE Access 2019; 7: 24122–24131.

16.

Saho

Uemura

Fujimoto M , et al. Evaluation of higher-level instrumental activities of daily living via micro-Doppler radar sensing of sit-to-stand-to-sit movement. IEEE J Transl Eng Health Med 2020; 8 (2100211): 1–11.

17.

Clarke

Lyketsos

, et al. Apathy and cognitive and functional decline in community-dwelling older adults: results from the Baltimore ECA longitudinal study. Int Psychoger 2010; 22: 819–829.

18.

Lam

Tam

Chiu

, et al. Depression and apathy affect functioning in community active subjects with questionable dementia and mild Alzheimer’s disease. Int J Geriatr Psychiatry 2007; 22: 431–437.

19.

Kojima

Iliffe

Taniguchi

, et al. Prevalence of frailty in Japan: a systematic review and meta-analysis. J Epidemiol 2017; 27(8): 347–353.

20.

Hölttä

Laakkonen

Laurila

, et al. Apathy: prevalence, associated factors, and prognostic value among frail, older inpatients. J Amer Med Direct Assoc 2012; 13: 541–545.

21.

Folstein

McHugh

PR.

Mini-mental state: a practical method for grading the cognitive state of patients for the clinician. J Psychiat Res 1975; 12: 189–198.

22.

Iwasa

Masui

Inagaki

, et al. Assessing competence at a higher level among older adults: development of the Japan science and technology agency index of competence (JST-IC). Aging Clin Exp Res 2018; 30(4): 383–393.

23.

DeLong

Clarke-Pearson

DL.

Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44(3): 837–845.

24.

Hilbe

JM.

Logistic regression models. Boca Raton, FL, USA: CRC Press, 2009.

25.

Fluss

Faraggi

Reiser

Estimation of the Youden Index and its associated cutoff point. Biometric J: J Math Meth in Biosci 2005; 47: 458–472.

26.

Hegeman

Shapkova

Honegger

, et al. Effect of age and height on trunk sway during stance and gait. J Vestib Res 2007; 17(2, 3): 75–87.

27.

Brady

Straight

Schmidt

, et al. Impact of body mass index on the relationship between muscle quality and physical function in older women. J Nutr Health Aging 2014; 18(4): 378–382.

28.

Sekhon

Launay

Chabot

, et al. Motoric cognitive risk syndrome: could it be defined through increased five-times-sit-to-stand test time, rather than slow walking speed? Front Aging Neurosci 2019; 10: 434.

29.

Hernandez

SSS

Vital

Garuffi

, et al. Apathy, cognitive function and motor function in Alzheimer’s disease. Dement Neuropsychol 2012; 6: 236–243.

30.

Moretti

Cavressi

Tomietto

Gait and apathy as relevant symptoms of subcortical vascular dementia. Am J Alzheimers Dis Other Demen 2015; 30: 390–399.

31.

Kanai

Kiyama

Goto

, et al. Use of the sit-to-stand task to evaluate motor function of older adults using telemetry. BMC Geriatr 2016; 16(1): 1–10.

32.

Fritz

Boileau

Stout

, et al. Relationships among apathy, health-related quality of life, and function in Huntington’s disease. J Neuropsychiatry Clin Neurosci 2018; 30: 194–201.

33.

Telenius

Engedal

Bergland

Effect of a high-intensity exercise program on physical function and mental health in nursing home residents with dementia: an assessor blinded randomized controlled trial. PLoS One 2015; 10(5): e0126102.

34.

Millor

Lecumberri

Gomez

, et al. Kinematic parameters to evaluate functional performance of sit-to-stand and stand-to-sit transitions using motion sensor devices: a systematic review. IEEE Trans Neural Syst Rehabil Eng 2014; 22(5): 926–936.

35.

Asker

Assal

Ding

, et al. Modeling of natural sit-to-stand movement based on minimum jerk criterion for natural-like assistance and rehabilitation. Adv Robotics 2017; 31: 901–917.