Toward Quantitative Neurology: Sensors to Assess Motor Deficits in Dementia

Abstract

Background:

Alzheimer’s disease (AD) is the most common neurodegenerative disorder which primarily involves memory and cognitive functions. It is increasingly recognized that motor involvement is also a common and significant aspect of AD, contributing to functional decline and profoundly impacting quality of life. Motor impairment, either at early or later stages of cognitive disorders, can be considered as a proxy measure of cognitive impairment, and technological devices can provide objective measures for both diagnosis and prognosis purposes. However, compared to other neurodegenerative disorders, the use of technological tools in neurocognitive disorders, including AD, is still in its infancy.

Objective:

This report aims to evaluate the role of technological devices in assessing motor involvement across the AD spectrum and in other dementing conditions, providing an overview of the existing devices that show promise in this area and exploring their clinical applications.

Methods:

The evaluation involves a review of the existing literature in the PubMed, Web of Science, Scopus, and Cochrane databases on the effectiveness of these technologies. 21 studies were identified and categorized as: wearable inertial sensors/IMU, console/kinect, gait analysis, tapping device, tablet/mobile, and computer.

Results:

We found several parameters, such as speed and stride length, that appear promising for detecting abnormal motor function in MCI or dementia. In addition, some studies have found correlations between these motor aspects and cognitive state.

Conclusions:

Clinical application of technological tools to assess motor function in people with cognitive impairments of a neurodegenerative nature, such as AD, may improve early detection and stratification of patients.

Keywords

Alzheimer’s disease cognitive impairment dementia mild cognitive impairment motor function sensors technology assessment

INTRODUCTION

Dementia is a progressive, irreversible syndrome characterized by cognitive decline in domains such as learning, memory, executive functions, language, attention, and perception together with functional impairment in daily activities. 1 According to the World Health Organization (WHO), more than 55 million people worldwide live with dementia, and 10 million new cases are reported each year. 2 Alzheimer’s disease (AD) is the more prevalent among the neurodegenerative disorder associated with dementia. To date the clinical course of AD is well characterized and its dementia phase is thought both to represent the final stage of a long-term condition and to be preceded by the well described mild cognitive impairment (MCI). Interestingly, motor signs have been increasingly described in subjects with MCI both as proxy markers of cognitive decline and as possible subtle signs of neurodegeneration (i.e., the mild motor impairment construct). Indeed, in recent years, researchers have pointed out the importance of individualized assessment and measurement tools that can capture both cognitive and motor aspects. 3 They also have outlined that in dementia and in preclinical or prodromal stages of neurocognitive disorders the motor aspects may also be involved early on in the disease course.4,5, 4,5

Motor changes (e.g., in balance, gait or hand movements) could thus be used as early, culture- and education-independent, and non-invasive markers of cognitive decline.6,7, 6,7 Indeed, to date there are several devices that can quantify motor parameters and detect impairments in an easy way both for patients and clinicians.8 –10 Intensifying research in this area could help physicians to thoroughly investigate the clinical status of the patient in the early stages of the disease in order to treat the symptoms as early as possible. Nevertheless, to date the tools available to clinicians are limited and generally do not include integrated observation of cognitive and sensorimotor aspects.

Various machine learning approaches are increasingly being applied in the diagnosis and progression of dementia, mostly in conjunction with neuroimaging data, including amyloid positron emission tomography (PET) imaging, magnetic resonance (MR) imaging, and combined PET and MR imaging.11,12, 11,12 However, the detection of early motor changes could be relevant for in depth investigation of the motor cognitive functions of patients. Indeed, in the last decades we have seen technological advances, which allowed us to push the boundaries of healthcare further. 13

Herein, we review various approaches in which technology is currently used in the assessment of motor aspects and their relationship with cognitive state in patients with MCI or dementia, with a focus on the AD spectrum. Our aims are to explore the potential benefits of technology in clinical assessment and diagnostic process, alongside providing an overview of the existing literature on motor state changes in the specified populations.

MATERIALS AND METHODS

This review has been conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations.14 –16 Based on the PICOS model, the “Participants” were patients with cognitive impairment of a neurodegenerative nature, such as AD; the “Intervention” was the use of technological tools among assessment methods, the “Comparator” was any conventional motor and cognitive evaluation method (e.g., paper and pencil neuropsychological tests); the “Outcome” was all measures related to cognitive state, gait and balance, physical functioning, disability, or Activities of Daily Living, and the “Study design” was controlled and open studies.

Selection criteria

Inclusion criteria were: 1) the presence of assessment through the use of technological tools as a typology of evaluation; 2) participants with cognitive impairment; 3) the application of conventional tests (e.g., paper and pencil neuropsychological tests); 4) all publication dates with no restrictions; 5) both controlled and open studies with case series were included.

Exclusion criteria were: 1) studies in which assessment does not concern motor aspects; 2) studies that monitored patients, rather than performing a single assessment, were excluded (e.g., tele monitoring, video monitoring, GPS etc.); 3) studies not written in English; 4) reviews, commentaries, abstracts, conference papers, studies on animal models or healthy subjects were excluded; 5) studies with pharmacologic goals; 6) patients with MCI/dementia caused by non-degenerative diseases, e.g., vascular diseases; 7) another cause with other neurodegenerative diseases such as Parkinson’s disease.

Study selection and data extraction

The outcomes of interest were related to motor function of upper and lower limbs, gait and balance, physical functioning, Activities of Daily Living with respect to cognitive function.

The PubMed, Web of Science, Scopus and Cochrane databases were searched on May 2023 for peer-reviewed with the following search string:

((Alzheimer’s* disease OR AD OR neurodegenerative disease OR Cognitive impairment OR Cognitive disorder OR Neurocognitive Disorders OR mild cognitive impairment OR MCI) AND ((motor function* OR Activities of Daily Living OR upper limb motor function OR posture OR balance OR lower limb motor function OR gait speed) AND (cognitive function OR cognitive impairment OR Cognitive disorder)) AND (wearable sensor* OR device OR sensor OR sensors OR inertial measurement OR Monitoring OR instrumentation* OR Wireless Technology* OR Wearable inertial sensors OR IMU OR accelerometer OR gyroscope OR magnetometer) AND (technology assessment OR sensors evaluation OR evaluation device)).

Search results were uploaded to Rayyan software, a web-based app to facilitate collaborations among reviewers during the selection of the studies. 17 Two authors (MH, SC) independently screened titles and abstracts. The reference lists of relevant papers were manually checked to identify additional studies potentially missed in the database search. Any disagreement was solved by consensus by a third reviewer (MP).

Authors independently extracted the following data from the included papers: study design, type of device, sample size, gender/age of patients, primary and secondary outcome measures, results.

Data collection and level of evidence

Variables of interest collected in this review included Author, Year, Type of study, Population, Device name, Intervention, Outcome Measures with device, Outcome Measures with the evaluation scale, Results, Level of evidence and grade of recommendation. The level of evidence of the studies was assigned according to the Oxford Centre for Evidence-Based Medicine framework (http://www.cebm.net/index.aspx?o=1025).

Quality assessment

The quality of evidence for each study was generally low as it was not possible to examine only prospective randomized controlled trials due to the heterogeneity of the devices and /or instruments under development, the small number of study participants, and the inadequate reporting of study details.

RESULTS

Selection of the studies

A total of 716 records were identified by means of literature search. After duplicates removal, 665 papers were screened through title and abstract and 128 papers were obtained for full-text screening that were independently reviewed and evaluated by two authors (MH, SC). Disagreement concerned 6 papers over the 128 selected for full-text examination (inter-rater agreement: 95.31% ) and was solved by consulting a third reviewer (MP). 21 articles fulfilled the inclusion criteria and were therefore included in the review (Fig. 1). The selected papers were grouped according to the investigated domain (i.e., gait and balance or upper limb performance) and type of evaluation device (i.e., wearable inertial sensors, tapping devices, tablet etc.).

Fig. 1

PRISMA diagram of the study (Page et al., 2021; www.prisma-statement.org).

Evaluation studies

Twenty one studies were aimed at the assessment of motor function in patients with cognitive deficits in different severity stages, with a focus on the AD spectrum (Table 1). If a paper also included other neurodegenerative diseases, it was included but we only considered participants who met the above-mentioned inclusion criteria for this review.

Table 1

Description of the studies included in the review

Domain	Typology of device	Author, Year	Type of study	Severity	Target Population	Device name	Intervention	Outcome Measures - device	Outcome Measures - scale	Results	Level of evidence - grade of recommendation
Gait and balance	Wearable inertial sensors / IMU	Buchman et al., 2020 19	Cohort study	MCI and dementia (M/F/age: N.S.) AD: 230 (M/F/age: N.S.) HC: 900	MCI: 119	Dynaport MT^®	N: 1249; 32 ft walk, TUG and standing with eyes close using a sensor attached on the lower back	12 gait scores from 26 parameters: sway, turn frequency, turn yaw, range, jerk, posterior, range, anterior-posterior, variability, regularity, cadence, speed	ADL, TUG, 19 cognitive tests for MCI/AD diagnosis (N.S.)	Almost all of the mobility metrics derived from an IMU are associated with the subsequent development of adverse health outcomes in older adults. Stride regularity, was found to be an independent predictor of MCI and AD, cadence of MCI and yaw of AD.	2b - B
		Bernhard et al., 2018 20	Cross-sectional study	Dementia	Dementia: 16 (M/F/age: N.S.)	Rehawatch^®	N: 16; gait and balance (three sensor-units at both ankles and lower back).	Sway area [cm2/s]; stride durations; stride length (m); number of steps (n); time (s)	TUG; muscle strength; muscle mass; FES-I; BDI-II; EQ-5D 5L; MMSE; TMT	Groups with ↓ gait speed and/or balance deficits showed impaired cognitive performance, supporting the association between motor performance and cognition.	3b - B
		Costa et al., 2016 21	Case-control study	Dementia	AD: 36 (M: 12; F: 24; age: 76±7) HC: 36 (M: 21; F: 15; age: 70±8)	/	N: 72; postural control kinematics from seven tasks (kinetic sensors and inertial unit attached to five body parts: trunk, legs and thighs)	total displacement on the transverse plane (cm); maximal displacement (cm); mean distance(cm); dispersion radius; maximal and mean linear velocity (cm/s); positioning (cm) on x-axis and y-axis; roll angle (degrees); pitch angle (degrees).	MoCA	Machine-learning classification systems can be useful in the diagnosis of AD, based on postural kinematics and MoCA. It was found a classification accuracy ranging between 91% and 96.6%	3b - B
		Mulas et al., 2021 22	Case-control study	Dementia	Young-old dementia: 28 (M: 9; F: 19; age: 71.3±2.9) Old-old dementia: 43 (M: 16; F: 27; age: 81.5±4.2) Young-old HC: 64 (M: 21; F: 43; age: 71.9±2.3) Old-old HC: 78 (M: 34; F: 44; age: 80.7±2.5)	G-Sensor^®	N: 213; 30 meters at a self-paced speed gait and TUG test with device (wearable inertial sensor attached to the trunk)	Gait (speed, stride length, cadence and duration of stance, swing and double support phase, walk ratio); TUG (duration, sit-to stand, intermediate 180° turn, final 180° turn, stand-to-sit)	ACE-R; TUG	Mobility alterations (↓ speed, stride length, cadence and swing phase duration, ↑ stance and double support duration, and altered coordination) in CI, especially in the old-old group. ↑ mobility alterations (especially for gait speed, stride length and TUG duration) correlates with the extent of cognitive impairment.	3b - B
		Hsu et al., 2014 23	Case-control study	Dementia	AD: 21 (M: 10; F: 11; age: 61.5±4.9) HC: 50 (M: 20; F: 30; age: 59.86±4.62)	/	N: 71; single and dual task 40 meters gait and balance tests (sensors attached to feet or to the waist)	Walking test (number of strides; walking time; stride length, frequency, speed, cadence, time; stance time; swing time; coefficient of variation of stride, stance and swing time; stance and swing period; coefficient of variation of stance and swing period); balance test (side by side stand with open or closed eye; left or right foot tandem stand with open or closed eye; stand on left or right foot)	MMSE, CASI	In single-task: in comparison with the HC, AD patients presented significantly ↓ stride length, stride speed, percentage of swing period; ↑ stance time, stride counts, speed, percentage of stance period, coefficient of variation of stand and swing period. In dual task: AD differed significantly from HC in all parameters and showed a significantly greater difference in gait parameters between single-task and dual-task conditions. In the balance tests, for medial-lateral direction, AD presented a significant difference with HC in the right foot tandem stand with eyes closed and stand on right foot.	3b - B
		Gillain et al., 2009 24	Case-control study	MCI and dementia	MCI:14 (M: 7; F: 7; age: 72.85) AD: 6 (M: 3; F: 3; age: 73.66) HC: 14 (M/F: N.S.; age: 73.53)	Locometrix^®	N: 34. single-leg balance single and dual task test, TUG single and dual task and pull test (parameters were recorded using the device)	Gait speed; stride frequency; stride length; stride regularity; stride symmetry	balance test, TUG, pull test, ADL, IADL, GDS, VAS; SHERPA; MMSE; DRS; FCSRT; ROCF; TAP	In single-task: AD showed ↑ TUG time and number of steps in comparison to MCI and HC; ↓ gait speed and stride length in comparison to HC; different regularity parameter in comparison to MCI. The MCI group had a ↓ gait frequency than HC. In dual-task: AD performed differently in all parameters but number of errors in comparison to HC and MCI; gait speed AD > MCI>HC; AD showed ↓ stride length and gait regularity than MCI and HC; MCI had ↓ stride frequency than HC. The device is able to define a specific gait profile for each cognitive profile. Gait speed and stride frequency seem correlated with cognitive level.	3b - B
		Zhou et al.,2022 25	Case-control study	MCI and dementia	CF: 41 (M:1; F: 44; age: 79.2±7.5) NCF: 80 (M:14; F: 66; age: 78.7±8.5)	LegSys	N: 111; single and dual task 10 meters gait. (5 sensors attached to lower back, left and right thighs and lower shins)	gait speed; stride length; gait cycle time; double support; gait speed unsteadiness; stride length unsteadiness	FES-I; CES-D; MMSE	In single task and dual task: CF had ↓ gait speed, stride length, ↑ gait cycle time, gait speed and stride length unsteadiness in comparison to NCF. In dual task only, double support was ↑ in the CF group. Gait metrics measured by sensors are potential digital biomarkers of CI with larger effects in dual tasks.	3b - B
		Huang et al, 2022 26	Cross-sectional study	MCI and dementia	Mild AD: 11; aMCI: 37 (M: 21; F: 27; age: 65.7±5.2) HC: 55 (M: 15; F: 40; age: 67.1±4.1)	JiBuEnTM	N: 103; 30 seconds free walking test wearing inertial sensors in shoe bottom, behind loin, the upper and lower limbs	Stride speed; cadence; heel strike angles	TUG, MMSE, AVLT, BNT, SCWT, SDMT, CDT	Memory deficit was associated with increased fall risk (↑ TUG time, ↓ heel strike angles, ↓ stride speed) and both with medial temporal, frontal, and parietal lobes atrophy	3b - B
		Kuan et al., 2021 27	Case-control study	MCI and dementia	aMCI: 30 (M: 12; F: 17; age: 74±6.7) Mild AD: 35 (M: 12; F: 13; age: 76±5) HC: 30 (M: 12; F: 17; age: 72.9±6.7)	Biodex Stability System &Physilog^®	N: 90 (MCI: 30; mild dementia: 30; HC: 30); posturography and balance and gait performance from self-paced single task walk longer than 30 meters	Sway indices; speed; stride length and variability	BBS, MMSE, CASI, grip force, calf circumference, Skeletal Muscle Index and Skeletal Muscle Mass	MCI had ↓ gait speed in comparison to HC; mild dementia group had ↓ stride length in comparison to HC; MCI and mild dementia showed ↓ BBS and ↑ indices of fall risk and sensory integration in comparison to HC.	3b - B
		Mishra et al., 2022 28	Case-control study	MCI	MCI: 11 (M: 5; F: 6; age: 83.7±2) HC: 14 (M: 10; F: 4; age: 69.3±2.1)	iTMT, LegSysTM	N: 25; using a screen in front of participant and moving the ankle of the dominant leg with a wearable sensor attached to the lower shin, participant had to move a cursor from circle “1” to circle “A” to circle “2” etc. as fast as possible	iTMT time	CES-D, MoCA, GAITRITE	↑ iTMT Time and ↓MoCA score in MCI in comparison to HC; iTMT time showed significant negative correlations with gait speed and MoCA.	4 - C
		Wiloth et al., 2016 29	Cohort study	Dementia	Dementia: 105 (M: 29; F: 76; age 82.7±5.9)	Physiomat	N: 99 (validity and feasibility); game tasks on Physiomat device (Physiomat Trail Making Tasks - PTMT; Physiomat balance tasks - PBT; the follow the ball test - FTBT)	Duration, sway path, sway area	GDS; MMSE; CERAD; FES-I; TUG; POMA; ZVT-G, ZN-G	Positive correlations between PTMT and TUG, FTBT, ZVT-G (temporal parameters), as well as PTMT and MMSE (total score). Negative correlations between PTMT and POMA, ZN-G (total score).	2b - B
		Mc Ardle et al., 2021 30	Case-control study	Dementia	AD: 32 (M: 15; F: 17; age: 77±6) HC: 25 (M: 11; F: 14; age: 74±9)	Axivity AX3	N: 57; sensor on lower back, six ten-meter walks at comfortable pace in lab	Pace (step velocity, step length, step time SD, wing time SD, stance time SD), variability (step length, step velocity SD), rhythm (step time, swing time, stance time), asymmetry (step time asymmetry, swing time asymmetry, stance time asymmetry), postural control (step length asymmetry)	CIRS-G; NART; BADLS; CDR; MMSE; ACE-III; GDS; ABC	AD showed ↑ step, stance and swing time, swing time SD, step length SD, and swing time asymmetry compared to HC.	3b - B
Console / Kinect	Yang et al., 2022 32	Cross-sectional study	Dementia	AD: 113 (M: 38; F: 75; age: 78.9±6.9)	Kinect Depth Sensor System, (Microsoft Corp.), PC	N: 113. Look straight and stand still for 40 s. 15 marked joints along the body.	20 angles from upper limbs; 10 angles from lower limbs.	MMSE, NPI, CASI, CDR	↓coordination function in the lower limbs, ↓MMSE, ↓CASI, ↑ CDR, and ↑ CDR-SB. ↓ coordination function upper limb,↑ CDR	3b - B
	Gait analysis	Montero-Odasso et al., 2009 18	Cohort study	MCI	MCI: 11 (M: 5; F: 6; age: 76.6±7.3)	GAITRite^® System	N: 11; electronic walkway under usual walking (single task) and counting backwards from 100 while walking (dual task)	gait velocity, step length, stride length, step time, stride time, and double support time	IADL; MMSE; MoCA	Step and stride time significantly ↑ under the dual task. Gait variability significantly ↑ under dual-task conditions in MCI.	4 - C
Upper limb performance	Tapping devices	Roalf et al., 2018 4	Case-control study	MCI and dementia	AD: 131 (M: 53; F: 78; age: 75.85±7.86) MCI: 46 (M: 19; F: 27; age: 72.3±9.03) HC: 62 (M: 19; F: 43; age: 71.37±9.47)	Light Beam Finger and Foot Tapper Test (NeuroCognitive Engineering)	N: 239. A light beam register tapping with index finger (of left or right hand). Three trials of 10 seconds for each hand	Number of taps; inter-tap interval (ITI); intra-individual variability (IIV)	MMSE, CERAD, CDR, DRS	AD and MCI showed ↓number of finger taps and ↑ ITI in comparison to HC. Number of finger taps and ITI were also significantly associated with MMSE performance.	3b - B
		Suzumura et al., 2016 33	Case-control study	MCI and dementia	MCI/AD: 23 (M/F: N.S.; age: 76.2) HC: 22 (M/F: N.S.; age: 63.1)	UB-1 (finger tapping device)	N: 27 (MCI/AD: 14; M/F: N.S.; age: 72.5±6.1; HC: 13; M/F: N.S.; age: 71.7±7.9). Four different thumb to finger-tapping tasks: with the dominant hand; with the dominant hand using a metronome; bimanual in-phase; bimanual anti-phase	44 parameters from: distance, velocity and acceleration amplitude, traveling and local distance, velocity, energy, number of freezing, velocity and acceleration of extending and flexing movements, contact duration, tapping interval, inter-tapping interval, similarity of hands, frequency and n° of taps.	MMSE, CDR	In bimanual tasks, MCI/AD ↓ total traveling distance, dispersion of the contact duration and rhythm perturbation compared to HC. Positive correlation: ↑ MMSE ↑ total traveling distance. Negative correlation: ↓ MMSE ↑ dispersion of the contact duration, rhythm perturbation, balance between flexing movement and extending movement and dispersion of inter-hand gaps.	4 - C
		Suzumura et al., 2021 34	Case-control study	MCI and dementia	AD: 49 (M/F: N.S.; age: 73.4±6.7); MCI: 20 (M: 11; F: 9; age: 76.7±4.2)	UB-2 (finger tapping device)	N: 64 (AD: 44; M: 19; F: 25; age: 73.8±7.0). Four different thumb to finger-tapping tasks: with a single hand (left or right hand), simultaneously with both hands, and alternate hands	36 parameters from: distance and acceleration amplitude, traveling distance, velocity, energy balance, number of freezing, velocity and acceleration of extending and flexing movements, contact duration, tapping interval, inter-tapping interval, inter-hand interval.	MMSE, Barthel Index	SD of distance rate of velocity peak in extending movement and SD of contact duration (hand tapping) were significantly ↑ in patients with AD or MCI and were associated with MMSE	3b - B
		Suzumura et al., 2022 35	Case-control study	MCI	MCI: 211 (M/F/age: N.S.) HC: 437 (M: 156; F: 281; age: 69.6±12.1)	UB-2 (finger tapping device)	N: 346 (MCI: 173; M: 74; F: 99; age: 77.2±6.8. HC: 173; M: 78; F: 95; age: 77.1±6.8). Four different thumb to finger-tapping tasks: with a single hand (left or right hand), simultaneously with both hands, and alternate hands	44 parameters from: distance, velocity and acceleration amplitude, traveling and local distance, velocity, energy, number of freezing, velocity and acceleration of extending and flexing movements, contact duration, tapping interval, inter-tapping interval (ITI), similarity of hands, frequency and n° of taps.	MMSE, Barthel Index, Grip	MCI showed ↓ number of taps, ↑ average of tapping interval, ↑ SD of ITI and ↑ n° of freezing calculated from acceleration during alternate tapping tasks with both hands in comparison to HC.	3b - B
	Tablet, mobile and PC	Holmes et al, 2022 36	Case-control study	MCI and dementia	CI: 38 (M: 23; F: 15; age: 73.6±6.4) Cognitively normal: 39 (M: 20; F: 19; age: 71.1±7.3)	Standardized laptop and smartphone	N: 77; a copying and a description task typing with mechanical keyboard and smartphone touch keyboard	Timing of press, release events in a typing stream, keyboard zone corresponding to each keystroke event, relative distance of the tap center to the target key center, key-content category, log of autocorrect and usage of word suggestions provided by the keyboard review tool and predictive engine, typing session context	ADLQ, DRS-II, FAB, MoCA	Moderate positive correlations between typing-based outcome and verbal memory, non-verbal memory, executive function. Weak positive correlation between typing-based outcome and language skills. Moderate positive correlation between the typing-based approach and MoCA.	3b - B
		Yan et al., 2008 37	Case-control study	MCI and dementia	MCI: 9 (M: 6; F: 3; age: 68.9±8.5) AD: 9 (M: 2; F: 7; age: 74.2±5.6) HC: 10 (M: 6; F: 4; age: 75.9±2.7)	Intuos2 USB WACOM^® digitizer; a hand-held stylus pen; MovAlyzeR v3.4 (computer program); laptop	N: 28. Move the stylus quickly between two dots. Four patterns: vertical; horizontal; forward slanted and backward slanted movements. 2 or 4 stroke (back and forth progressions) conditions.	Movement time and jerk	MMSE	HC 30% faster than MCI; MCI 13% faster than AD. ↓ speed and ↓ jerky movement execution might be associated with ↑cognitive dysfunction in AD and MCI, therefore ↓ MMSE.	4 - C
	Wearables / IMU	Rovini et al., 2021 38	Case-control study	MCI	MND: 17 (M: 4; F: 13; age: 77.1±5.15) HC: 10 (M: 4; F: 6; age: 63,7±9,9)	SensRing	N: 18 (MND: 8); reach-to-grasp and after-grasp sequences with a can (one sensor in the proximal phalanx of the dominant index finger)	12 kinematic parameters	MMSE	Positive correlation between ↑ MMSE and ↑ peak velocity. Negative correlation between ↑ MMSE and ↓ deceleration time, ↓variability of the movements.	4 - C

ABC, Activity Balance Confidence Scale; ACE-III, Addenbrookes Cognitive Examination III; ACE-R, Addenbrookes Cognitive Examination Revised; AD, Alzheimer’s Disease; ADL, Activity of Daily Living; ADLQ, Activity of Daily Living Questionnaire; AD-MCI, MCI/Alzheimer’s disease; aMCI, amnestic Mild Cognitive Impairment; AVLT, Auditory Verbal Learning Test; BADLS, Bristol Activities of Daily Living Scale; BBS, Berg Balance Scale; BDI-II, Beck’s Depression Inventory II; BNT, Boston Naming Test; CASI, Cognitive Assessment Screening Instrument; CDR, Clinical Dementia Rating Scale; CDR-SB, Clinical Dementia Rating Scale – sum of boxes CDT, Clock Drawing Test; CERAD, Consortium to Establish a Registry for Alzheimer’s Disease; CES-D, Center for Epidemiologic Studies-Depression; CF, cognitive frailty group; CI, cognitively impaired; CIRS-G, Cumulative illness rating scale - Geriatrics; DLB, Dementia with Lewy bodies; DLQ, Daily Living Questionnaire; DRS, Dementia Rating Scale; DRS-II, Dementia Rating Scale 2; EQ-5D 5L, EuroQol – 5D 5 level; F, female; FAB, Frontal Assessment Battery; FCSRT, Free and Cued Selective Reminding Test; FES-I, Falls Efficacy Scale-International; FTBT, Follow-the-Ball Task; GDS, Geriatric Depression Scale; GDS-15, Geriatric Depression Scale 15-Item; HC, healthy control group; IADL, Instrumental Activity of Daily Living; IMU, inertial measurement unit; iTMT, instrumented trail-making task; M, male; MCI, Mild Cognitive Impairment; MMSE, Mini-Mental State Examination; MNA, Mini Nutritional Assessment; MND, Mild Neurocognitive Disorder; MoCA, Montreal Cognitive Assessment; N, number; N.S., not specified; NART, National Adult Reading Test; NCF, non-cognitive frailty group; ND, Neurodegenerative Disease; NPI, Neuropsychiatric Inventory; PC, Personal computer; POMA, Performance-Oriented Mobility Assessment; RCT, randomized clinical study; ROCF, Rey– Osterrieth Complex Figure; SD, standard deviation; SCWT, Stroop Color-Word Test; SDMT, Symbol Digit Modalities Test; SHERPA, Score Hospitalier d’évaluation du risque de perte d’autonomie; TAP, Test for Attentional Performance; TMT, Trail Making Test; TUG, Timed-Up and Go; VAS, Visual analogue scale; ZN-G, repeating numbers; ZVT-G, modified version of the Trail-Making-Test A.

Participants

The total number of participants is 3418 (controls = 1926) with a sample size ranging from 11 18 to 1249 19 and a median of 72 participants. The total of patients involved in the studies is 1492 (range = 11– 349) with a median of 38. Given the considerable heterogeneity in the approach to diagnostic labelling used in the reported papers, the target population was reported as MCI in 4 articles [250 patients (16.8% ), median = 14, range = 11–211], as well as AD [202 patients (13.5%), median = 34, range = 21–113], while MCI and AD were reported in 9 studies [807 patients (54.1%), median = 48, range = 18–349]. The remainder 4 reported on people with unspecified dementia [233 patients (15.6%), median = 56, range = 16–105]. The intervention was applied on 3063 subjects (median = 72, range = 11– 1249).

Outcome measures: Devices used in the studies

Wearable inertial sensors to evaluate gait and balance

Twelve studies evaluated MCI or dementia patients using wearable inertial sensors.19 –30 These types of instruments utilize different inertial sensors, such as accelerometers and gyroscopes, to detect and quantify motion along the axes. In some cases, these sensors are worn by patients, for instance, at the waist or ankle level, while in other cases, they are integrated into larger devices, such as an unstable platform.

In a cross-sectional study, Bernhard et al. 20 tested the gait and balance with three sensor-units at both ankles and lower back. The study designed for unspecified dementia patients including walked, static balance with 4 positions of the feet (open, closed, semi-tandem and tandem stance), then performed on a foam pad, and found groups with gait speed and/or balance deficits showed impaired cognitive performance, supporting the association between motor performance and cognition. In another case-control study, Costa et al. 21 investigated an intervention using a combination of kinetic sensors and inertial units (trunk, legs and thighs) to assess the kinematics of postural control in seven tasks in patients with AD. After recognizing the interaction between postural stability and cognitive performance, they found that the analysis of postural kinematics has the potential to be used as a complementary biomarker in the diagnosis of AD. In a case-control study on individuals with unspecified dementia Mulas et al. 22 applied a wearable inertial sensor to the trunk of the participants during a natural 30 meters straight walk at a self-paced speed and Timed Up and Go (TUG) test. Gait parameters such as speed, stride length, cadence and duration of stance, swing and double support phase and walk ratio were obtained from the natural walk and 5 other parameters came from Timed Up and Go (duration, sit-to stand, intermediate 180° turn, final 180° turn, stand-to-sit). They found that aspects of mobility were significantly impaired in the presence of a cognitive deficit in comparison to healthy controls. Gait speed, stride length and Timed Up and Go duration are correlated with the extent of cognitive impairment. Buchman et al. 19 performed a longitudinal cohort study to explore the mobility of MCI and AD subjects with a wearable sensor positioned on the lower back with a neoprene belt. They extracted 26 parameters from 32 ft walk, Timed Up and Go and standing with eyes closed (one of which with 3 subtasks) and found that 10 out of 12 gait scores (specified in Table 1) were associated with incident mortality, ADL disability, motor disability, incident MCI or incident AD. The study findings suggest that modelling combinations of different mobility measurements are differentially associated with specific health outcomes. Hsu et al. 23 conducted a case-control study in patients with AD to investigate the use of foot-mounted sensors in a 40-meter walk test, which consisted of a single and dual task, and a balance test with waist-mounted sensors. The gait and balance measurement device consisted of a triaxial accelerometer, a uniaxial and biaxial gyroscope, a microcontroller and a micro-SD flash memory card. It considered 16 parameters from the gait test and 16 parameters from the balance test. The device is a convenient tool for automatically analyzing gait and balance information. In AD patients, the results showed a statistically significant difference with the healthy control group in gait parameters both in and between single-task and dual-task and a significantly higher average sway speed in the balance tests. In a case-control study, Gillain et al. 24 tested the efficacy of the tri-axial accelerometer to evaluate body movements in all three spatial planes (the gait performance with the pull test, the single-leg balance test, the Timed Up and Go test and a 30-meter stable gait test) in MCI, AD and Healthy Control. They showed the clinical motor tests were not able to significantly differentiate between groups, while the device demonstrated significantly lower gait speed and stride length regularity in AD patients. Furthermore, the MCI patients had a lower stride frequency in comparison to Healthy Control. Gait speed and stride frequency positively correlated with the patient’s global cognitive level. Zhou et al. 25 reported a case-control study that explored the effect of five wearable sensors attached to the lower back, left and right thighs and lower shins to quantify speed, stride length, cycle time and double support during gait task in patients with amnestic MCI and mild dementia and found gait parameters are potential digital biomarkers of cognitive impairment. Huang et al. 26 conducted a cross-sectional study in an amnestic MCI and mild AD sample to explore 30 s free walking test wearing inertial sensors to measure stride speed, cadence and heel strike angles. They found a significant association between memory deficit and increased fall risk and this latter aspect might represent an objective marker for the monitoring of memory and fall risk itself. Kuan et al. 27 assessed posturography, through the use of the stability system, and quantified gait performances wearing a device with inertial sensors on healthy controls, amnestic MCI and mild dementia due to AD patients. Indices of sensory integration were significantly worse in mild dementia than in MCI and Healthy Control. Clinical motor (Berg Balance Scale) and fall risk tests significantly differed between MCI and mild dementia in comparison to healthy controls. Moreover, results showed significantly lower different speed scores in MCI and lower stride length in mild dementia groups in comparison to healthy controls. Mishra et al. 28 evaluated the association of instrumented Trail-Making Task (iTMT), an interactive game-based software, to calculate instrumented Trail-Making Task time in patients with MCI. They found the device is able to distinguish individuals with MCI from Healthy Control older adults. Furthermore, they found correlations between instrumented Trail-Making Task, Montreal Cognitive Assessment (MoCA) score and gait speed. Wiloth and colleagues 29 validated a game-based assessment of balance-cognitive abilities, using a moveable plate connected to a computer and monitor, in people with unspecified mild-to-moderate dementia. They applied three tasks: Physiomat Trail Making Tasks (PTMT), an instrumental Trail Making Tasks; Follow The Ball Test (FTBT), in which participants had to follow a moving ball on the screen using balance; Physiomat Balance Tasks (PBT), in which participants had to remain as still as possible. The results of this cohort study showed a correlation between the time taken to complete Physiomat Trail Making Tasks and cognitive and motor tests. In a case-control study by Mc Ardle et al. 30 several aspects of gait were assessed using a sensor placed on the lower back of healthy participants and patients with AD performing a ten-meter walk at comfortable pace in the lab. AD patients were more variable for step, stance and swing time, spent longer in swing time and had greater swing time asymmetry in comparison to controls.

Considering these 12 studies on the use of wearable inertial sensors or Inertial Measurement Unit (IMU) for gait and balance, seven studies reported that slower gait speed was associated with poorer global cognition.20,22–27 , 20,22–27 Six studies showed that another important parameter is stride, especially stride length, which is lower in patients compared to healthy controls.19,22–25,27 , 19,22–25,27 Three studies found differences in the swing aspects of gait in patients compared to controls: Mulas reported a decrease in swing phase duration, Hsu showed a decrease in swing period percentage and McArdle highlighted an increase in swing time, swing time asymmetry and standard deviation.22,23,30 , 22,23,30 Finally, two studies reported a decrease in cadence in patients compared to controls.19,22, 19,22

In four studies, dual task paradigm - motor and cognitive - enhanced differences between the groups.18,23–25 , 18,23–25 Globally, gait and balance performances are reported as worse in patients in comparison to healthy controls and emerges an association between motor abilities and cognition.

Gaming consoles and kinect to evaluate gait and balance

Gaming consoles are already being used in physiotherapy rehabilitation and have spawned a whole new area— exergames— that allows people to carry out rehabilitation programs in a stimulating and interactive way. 31 In addition to this approach, this field also has the potential to measure movement for purposes other than rehabilitation. We report a cross-sectional study with people suffering from AD. Yang et al. 32 used the Kinect Depth Sensor System, which is part of the Microsoft Xbox console ecosystem, to measure a total of thirty angles formed by the limbs from fifteen different body joints with the centerline. This was done without the participants having to perform any task: they simply had to stand still and look straight ahead for forty seconds. In conclusion, authors found that changes in coordination function in limbs can reflect the clinical outcomes, cognition and global function.

Gait cycle formal analysis

Another aspect assessed through the use of technology is the patient’s gait cycle, which is composed of the stance phase and the swing phase, both of which are subdivided into several other phases. Of the twenty-one papers included in this review, one addressed this topic.

Montero-Odasso and colleagues 18 analyzed gait parameters in a cohort study of MCI patients walking on an engineered mat (GAITrite) during single and dual cognitive tasks. They found that the dual task, which involved counting backwards from 100, played a role in the patients’ gait performance, leading to an increase of step and stride time compared to the single task.

Finger tapping devices to assess upper limb function

Four studies used finger tapping to study motor performances of AD or MCI patients. These finger tapping sensors devices are typically placed on the fingers and are capable of detecting and measuring the intensity, frequency, duration, and pattern of taps produced by the individual. They can be used in research settings or clinical assessments to objectively measure hand performance, compare motor parameters and detect any subtle changes in motor function.

In a case-control study in MCI, AD and Healthy Control Suzumura et al. 33 applied a finger tapping device (UB-1) and found a decrease in patients’ performances in total travelling distance, dispersion of the contact duration and rhythm perturbation compared to Healthy Control. Moreover, they found a positive correlation between Mini-Mental State Examination (MMSE) and total travelling distance and a negative correlation between MMSE and dispersion of the contact duration, rhythm perturbation, balance between flexing movement and extending movement and dispersion of inter-hand gaps. Thus, finger dexterity is associated with decline in cognitive function evaluated with MMSE and its assessment can be used as an indicator of cognitive function. Suzumura and colleagues 34 requested to wear a device with magnetic sensors (UB-2) on their thumbs and index fingers of both hands and to perform finger tapping movements in four different scenarios. They collected 36 different parameters from the instrument, along with MMSE and Barthel Index scores. Results showed significant associations between the standard deviation (SD) of distance rate of velocity peak in extending movement and the SD of contact duration and cognitive function. In a case-control study, Suzumura et al. 35 used the same finger tapping device of the previous study 34 to measure 44 parameters, such as distance, velocity, acceleration, energy, number of freezing, contact duration, tapping interval, inter-tapping interval, similarity of hands, frequency and number of taps, in patients with MCI and Healthy Control. They reported a significant decrease in the number of taps and an increase in the average of tapping interval, inter-tapping interval and number of freezing in patients. Roalf et al. 4 applied a finger and foot tapping test in a case-control study on AD and MCI and found a decrease in the number of finger taps and an increase of inter tapping interval in AD and MCI in comparison to Healthy Control, with a unique pattern of dysfunction for each disorder, and these two parameters were associated with MMSE.

Globally, the most evaluated parameters are: inter tapping interval,4,35, 4,35 number of taps,4,35, 4,35 and contact duration.33,34, 33,34

Touchscreen-based tools to assess upper limb function

Given their widespread diffusion and the increasing ease of use facilitated by touchscreen, voice commands and artificial intelligence, tablets, smartphones and Personal Computers serve as valuable tools for assessing motor or cognitive performances in general. In this review, we found two papers using this type of device.

Holmes and colleagues 36 involved people with MCI, MCI/AD and AD in two tasks among two paradigms: copying an excerpt and describing a picture with the mechanical keyboard of a laptop; transcribing a different passage and answering three questions with the touch keyboard of a smartphone. Using machine learning models, they found a significant correlation between users’ patterns of typing on devices and their verbal memory, non-verbal memory, executive functions and language Furthermore, their method was able to produce quantitative measures of both global cognition and its relative subdomains directly from the analysis of the mentioned tasks.

Secondly, Yan et al. 37 conducted a case-control study in order to quantify, using technological devices, fine motor control of people with a diagnosis of MCI or probable AD. Specifically, participants were instructed to move between two dots as fast as possible using a stylus pen on a digitizer in two or four stroke conditions. The healthy controls group resulted to be faster, smoother and steadier than MCI group, which, in turn, performed better in each of these aspects than AD group.

As the instruments used in the two studies are very different, the results are not comparable. Nonetheless, speed is an important element in Yan et al. 37 , in line with other studies mentioned above dealing with different instruments, such as in the previous chapters “Wearable devices/IMU” and “Tapping device”. Furthermore, these two studies show that it is possible to distinguish cognitively impaired people and healthy controls by analyzing simple tasks performed with the hands, such as typing on a keyboard or using a pen on a tablet.

Wearable inertial sensors to assess upper limb function

Rovini et al. 38 performed an analysis of a case-control study in dysexecutive mild neurocognitive disorder (MND) patients to measure kinematic parameters with one ring-shaped sensor in the proximal phalanx of the dominant index finger. They found that cognitively impaired subjects exhibited slower movement, slower reaction and longer deceleration and had more difficulties in the modulation of the kinematics of their movements in comparison to Healthy Control. The device is small and might have the potential to be used for clinical application or for remote monitoring at home.

Outcome measures: Clinical scales used in the selected studies

We observed a significant heterogeneity in the scales used to assess the clinical correlates of the motor impairment observed in the selected studies. Table 2 facilitates a comparison of cognitive, psychological and functional abilities (Activities of Daily Living, or ADL) assessment tools used in the selected articles. As indicated in the Table 2, assessment tools may evaluate different stages of disease severity, domains or aspects of a particular domain. In some studies, multiple clinical tests were used and functional information from motor scales was included to provide a better understanding of the patient. Regarding the assessment of motor ability, studies generally used scales concerning stability and/or the fall risk. Some studies used scale to assess body composition or strength and physical impairment (Table 3).

Table 2

Cognitive and psychological instruments applied in the studies included in the review

Test	References	COGNITIVE DOMAIN						Neuropsychiatric	Quality of life	Sleep quality and
		Orientation	Attention - Working Memory	Memory	Visuospatial	Language	Executive Functions	Symptoms		disturbances
ACE-III	30	✓	✓	✓	✓	✓
ACE-R	22	✓	✓	✓	✓	✓
AVLT	26			✓
BDI-II	20							✓
BNT	26					✓
CASI	23,27,32 , 23,27,32	✓	✓	✓	✓	✓	✓
CDR	4,30,32,33 , 4,30,32,33	✓		✓			✓
CDT	26				✓		✓
CERAD	4,29, 4,29	✓	✓	✓	✓	✓	✓
CES-D	25,28, 25,28							✓
DRS	4,24, 4,24	✓	✓	✓	✓		✓
DRS-II	36	✓	✓	✓	✓		✓
EQ-5D-5L	20							✓
FAB	36		✓				✓
FCSRT	24			✓
GDS	24,29,30 , 24,29,30							✓
IADL	18,24, 18,24
MMSE	4,18,20,23–27,29,30,32–35,37,38 , 4,18,20,23–27,29,30,32–35,37,38	✓	✓	✓	✓	✓
MoCA	18,21,28,36 , 18,21,28,36	✓	✓	✓	✓	✓	✓
NART	30					✓
NPI	32							✓		✓
ROCF	24			✓	✓
SCWT	26						✓
SDMT	26		✓				✓
SHERPA	24	✓	✓	✓
TAP	24		✓				✓
TMT	20		✓				✓
VAS	24								✓
ZN-G	29		✓
ZVT-G	29		✓

ACE-III, Addenbrookes Cognitive Examination III; ACE-R, Addenbrookes Cognitive Examination Revised; AVLT, Auditory Verbal Learning Test; BDI-II, Beck’s Depression Inventory II; BNT, Boston Naming Test; CASI, Cognitive Assessment Screening Instrument; CDR, Clinical Dementia Rating Scale; CDT, Clock Drawing Test; CERAD, Consortium to Establish a Registry for Alzheimer’s Disease; CES-D, Center for Epidemiologic Studies-Depression; CTP, Cookie Theft Picture; CVLT-SF, California Verbal Language Test-II Short Form; DRS, Dementia Rating Scale; DRS-2, Dementia Rating Scale 2; EQ-5D 5L, European Quality of Life 5 Dimensions 5 Level Version; FAB, Frontal Assessment Battery; FCSRT, Free and Cued Selective Reminding Test; GDS, Geriatric Depression Scale; IADL, Instrumental Activity of Daily Living; JLO, Judgment of Line Orientation; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; NART, National Adult Reading Test; NPI, Neuropsychiatric Inventory; PSQI, Pittsburgh Sleep Quality Index; ROCF, Rey– Osterrieth Complex Figure; SCWT, Stroop Color-Word Test; SDMT, Symbol Digit Modalities Test; SHERPA, Score Hospitalier d’évaluation du risque de perte d’autonomie; TAP, Test for Attentional Performance; TCF, Taylor Complex Figure; TMT, Trail Making Test; VAS, Visual Analogue Scale; ZN-G, repeating numbers resembling Digit span task; ZVT-G, modified version of the Trail-Making-Test A.

Table 3

Motor and functional instruments applied in the studies included in the review

Test	Reference	Stability	Risk of fall	Body composition-strength	Physical impairment	Functional domain (ADL)	Functional domain (IADL)
ABC	30	✓	✓
ADL	19,24, 19,24					✓
ADLQ	36					✓	✓
BADLS	30					✓	✓
Barthel Index	34,35, 34,35					✓	✓
BBS	27	✓	✓
CIRS-G	30				✓
FES-I	20,25,29 , 20,25,29		✓
IADL	18,24, 18,24
Largest circumference of the calf	27			✓
Single-leg balance test	27	✓
Pull test	27	✓
Skeletal muscle index	27			✓
Strength and muscle mass	20,27, 20,27			✓
POMA	29	✓	✓
TUG	19,20,22,24,26,29 , 19,20,22,24,26,29		✓

ABC, Activities-specific Balance Confidence; ADL, Activity of Daily Living; ADLQ, Activity of Daily Living Questionnaire; BADLS, Bristol Activities of Daily Living Scale; BBS, Berg Balance Scale; CIRS-G, Cumulative Illness Rating Scale; FES-I, Falls Efficacy Scale-International; POMA, Performance-Oriented Mobility Assessment; Tinetti; TUG, Timed Up and Go Test; VAS, Visual Analogue Scale.

DISCUSSION

We provide a review on motor function parameters evaluated with devices or sensors in dementia and/or MCI patients. We included 21 articles with heterogeneous devices, study outcomes and study design. Some studies focused on balance, gate and generally lower limbs (N = 14) and the others used the technologies in upper limbs (N = 7).

We found 13 studies on IMU sensors (12 lower limbs, 1 upper limb) and that have shown a potential efficacy in assessing motor involvement. They provide a continuous registration of motor parameters and can detect subtle changes in motor performance, including gait and balance, in a wide spectrum of settings. These devices are often lightweight and easy to use. Our finding of different protocols with a highly heterogeneous technology applied showed that challenges exist in validating and standardizing the data collected by these sensors across different populations. However, it seems that speed for lower and upper limb (N = 8) and stride for lower limb (N = 6) are two important parameters during the evaluation with wearable sensors. Further research is needed to establish the accuracy, reliability, and clinical utility of wearable sensors in assessing motor involvement in AD.

Two studies used motion systems to analyze movement patterns: in one case, authors used a sensorized mat, Gaitrite, and in the other one, they used a camera system, Kinect. The devices offer high-resolution assessment of motor function by capturing and analyzing movement patterns. While Gaitrite analysis focuses more on stride and gait velocity as the parameters to be mainly considered, like the studies with IMU sensors, Kinect focuses on movement coordination. However, motion systems are typically expensive and require specialized equipment and expertise for setup and analysis.

Six papers explored tapping sensors, tablet/smartphone or personal computer protocols, where they evaluated coordination and speed during the hand and fingers movement. The studies showed these tasks can be used for the evaluation of hand function and motor skills. Fine hand movements tests can be useful in assessing various motor aspects, such as finger dexterity, coordination, speed, and rhythm. They can provide quantitative data on the individual’s motor performance, allowing an objective evaluation. Three underlined speeds (expressed as number of taps in finger tapping paradigms) as a core parameter to take into consideration as a primary outcome.

Nonetheless, these apparently promising results should be cautiously considered, because of several potential confounders. First, target population, severity and devices differed across studies. As noted, motor symptoms in dementia patients can vary depending on the type and stage of cognitive decline. Therefore, it is important to evaluate changes in motor function in early stages, such as MCI, and possibly to stratify patients according to the different underlying neuropathology. Second, methodological differences, including type of movement, position of the sensors, registering parameters and outcome measures may hamper the findings replication. Third, the reported studies did not identify a “gold-standard” battery for a comprehensive assessment of motor function in subjects with AD and other dementing conditions. This is confirmed by the inhomogeneity of motor tests applied in the selection of studies reported in this review in Table 3. There is also a lack of agreement about a shared battery of cognitive tests to apply to explore the cognitive correlated of motor impairments even though the two major screening tests, MMSE and MoCA, are the most frequently used in the review, as reported in section “Outcome measure - Scale”. These tests cannot provide a comprehensive view of a patient’s cognition, although their main advantages are that they are universally recognized as global indicators of cognition and that they are quick and easy to administer. 39 However, regarding the rating scales, our findings are in keeping with previously published review on Outcome Measures in Dementia and MCI Trials, confirming the fact that evaluation scales in dementia/MCI trials remain inhomogeneous and diverse. 40

The evaluation of motor function with the device revealed that gait, balance,18–26,28–30,32 , 18–26,28–30,32 and upper limbs movements4,27,33–37 , 4,27,33–37 were associated with cognitive function, confirming a significant association between cognitive function and changes in motor abilities. Furthermore, traditional motor scales are usually not sensitive enough to detect small early changes in patients with cognitive impairment of a neurodegenerative nature, such as AD. We believe that the application of a blended assessment, consisting of traditional tests and modern evaluation methods with technological devices, might represent a better option for clinicians in order to differentiate between pathological and preserved motor function. Instruments should be valid, reliable and responsive to provide clinically important changes over time. Research efforts are necessary in order to validate existing devices and to identify the best performing ones, with the aim to identify the subtle motor disturbances that can advise the onset of the occurrence of a cognitive failure.

Limitation of the study

There was a limited possibility to perform a systematic review because of the different outcomes measured across studies and quality of the papers. Larger and more detailed device studies are needed to further explore their effectiveness given the reliable results of the limited current literature.

AUTHOR CONTRIBUTIONS

Mehrnaz Hamedani (Conceptualization; Data curation; Formal analysis; Methodology; Writing – original draft; Writing – review & editing); Stefano Caneva (Conceptualization; Data curation; Formal analysis; Methodology; Writing – original draft; Writing – review & editing); Gian Luigi Mancardi (Methodology; Writing – review & editing); Paolo Alessandro Alì (Writing – review & editing); Pietro Fiaschi (Methodology; Writing – review & editing); Federico Massa (Writing – review & editing); Angelo Schenone (Funding acquisition; Methodology; Writing – review & editing); Matteo Pardini (Conceptualization; Funding acquisition; Methodology; Supervision; Writing – original draft; Writing – review & editing).

Footnotes

ACKNOWLEDGMENTS

AS gratefully acknowledges the support of the project “RAISE - Robotics and AI for Socioeconomic Empowerment” funded by the European Union - NextGenerationEU. MH and FM acknowledge the support of #NEXTGENERATIONEU (NGEU), funded by the Ministry of University and Research (MUR), National Recovery and Resilience Plan (NRRP), project MNESYS (PE0000006) – A Multiscale integrated approach to the study of the nervous system in health and disease (DN. 1553 11.10.2022). MP thanks the Italian Ministry of Health for funding the project “Dalla neurologia quantitativa alla tele-neurologia: quali strumenti per l’assessment computerizzato della disabilità motoria?”. 5×1000 2018-2019. We also thank the Department of Neurosciences, Rehabilitation, Ophthalmology, Genetic and Maternal Infantile Sciences (DINOGMI), for the valuable collaboration on this article.

The views and opinions expressed are those of the authors alone and do not necessarily reflect those of the European Union or the European Commission. Neither the European Union nor the European Commission can be held responsible for them.

FUNDING

This study was supported by grants from the Italian Ministry of Health, 5×1000 and European Union – NextGenerationEU.

CONFLICT OF INTEREST

The authors have no conflicts of interest to report.

DATA AVAILABILITY

The data supporting the results of this study are openly available online.

References

Ramsey

, Arnold

. Chapter 11 - Prognostication in dementia. In: Miyasaki

, Kluger

(eds) Handbook of Clinical Neurology. Elsevier, pp. 163–174.

Shin

J-H

. Dementia epidemiology fact sheet 2022. Ann Rehabil Med 2022; 46: 53–59.

Bahureksa

, Najafi

, Saleh

, et al. The impact of mild cognitive impairment on gait and balance: a systematic review and meta-analysis of studies using instrumented assessment. Gerontology 2017; 63: 67–83.

Roalf

, Rupert

, Mechanic-Hamilton

, et al. Quantitative assessment of finger tapping characteristics in mild cognitive impairment, Alzheimer’s disease, and Parkinson’s disease. J Neurol 2018; 265: 1365–1375.

Gillain

, Dramé

, Lekeu

, et al. Gait speed or gait variability, which one to use as a marker of risk to develop Alzheimer disease? A pilot study. Aging Clin Exp Res 2016; 28: 249–255.

Albers

, Gilmore

, Kaye

, et al. At the interface of sensory and motor dysfunctions and Alzheimer’s disease. Alzheimers Dement 2015; 11: 70–98.

Kueper

, Lizotte

, Montero-Odasso

, et al. Cognition and motor function: The gait and cognition pooled index. PLoS One 2020; 15: e0238690.

Heldman

, Espay

, LeWitt

, et al. Clinician versus machine: reliability and responsiveness of motor endpoints in Parkinson’s disease. Parkinsonism Relat Disord 2014; 20: 590–595.

Matias

, Paixão

, Bouça

, et al. A perspective on wearable sensor measurements and data science for Parkinson’s disease. Front Neurol 2017; 8: 677.

10.

FitzGerald

, Lu

, Jareonsettasin

, et al. Quantifying motor impairment in movement disorders. Front Neurosci 2018; 12: 202.

11.

Wang

, Sheng

, Xu

, et al. Develop a diagnostic tool for dementia using machine learning and non-imaging features. Front Aging Neurosci 2022; 14: 945274.

12.

, Jiang

, Wang

, et al. Applied machine learning in Alzheimer’s disease research: omics, imaging, and clinical data. Emerg Top Life Sci 2021; 5: 765–777.

13.

Flessa

, Huebner

. Innovations in health care—a conceptual framework. Int J Environ Res Public Health 2021; 18: 10026.

14.

Moher

, Shamseer

, Clarke

, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Syst Rev 2015; 4: 1.

15.

Liberati

, Altman

, Tetzlaff

, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. J Clin Epidemiol 2009; 62: e1–34.

16.

Page

, McKenzie

, Bossuyt

, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021; 372: n71.

17.

Ouzzani

, Hammady

, Fedorowicz

, et al. Rayyan—a web and mobile app for systematic reviews. Syst Rev 2016; 5: 210.

18.

Montero-Odasso

, Casas

, Hansen

, et al. Quantitative gait analysis under dual-task in older people with mild cognitive impairment: a reliability study. J NeuroEngineering Rehabil 2009; 6: 35.

19.

Buchman

, Dawe

, Leurgans

, et al. Different combinations of mobility metrics derived from a wearable sensor are associated with distinct health outcomes in older adults. J Gerontol A Biol Sci Med Sci 2020; 75: 1176–1183.

20.

Bernhard

, Sartor

, Bettecken

, et al. Wearables for gait and balance assessment in the neurological ward - study design and first results of a prospective cross-sectional feasibility study with 384 inpatients. BMC Neurol 2018; 18: 114.

21.

Costa

, Gago

, Yelshyna

, et al. Application of machine learning in postural control kinematics for the diagnosis of Alzheimer’s disease. Comput Intell Neurosci 2016; 2016: e3891253.

22.

Mulas

, Putzu

, Asoni

, et al. Clinical assessment of gait and functional mobility in Italian healthy and cognitively impaired older persons using wearable inertial sensors. Aging Clin Exp Res 2021; 33: 1853–1864.

23.

Hsu

Y-L

, Chung

P-CJ

, Wang

W-H

, et al. Gait and balance analysis for patients with Alzheimer’s disease using an inertial-sensor-based wearable instrument. IEEE J Biomed Health Inform 2014; 18: 1822–1830.

24.

Gillain

, Warzee

, Lekeu

, et al. The value of instrumental gait analysis in elderly healthy, MCI or Alzheimer’s disease subjects and a comparison with other clinical tests used in single and dual-task conditions. Ann Phys Rehabil Med 2009; 52: 453–474.

25.

Zhou

, Park

, Shahbazi

, et al. Digital biomarkers of cognitive frailty: the value of detailed gait assessment beyond gait speed. Gerontology 2022; 68: 224–233.

26.

Huang

, Zhou

, Liu

, et al. High fall risk associated with memory deficit and brain lobes atrophy among elderly with amnestic mild cognitive impairment and mild Alzheimer’s disease. Front Neurosci 2022; 16: 896437.

27.

Kuan

Y-C

, Huang

L-K

, Wang

Y-H

, et al. Balance and gait performance in older adults with early-stage cognitive impairment. Eur J Phys Rehabil Med 2021; 57: 560–567.

28.

Mishra

, Park

, Zhou

, et al. Evaluation of motor and cognitive performance in people with Parkinson’s disease using Instrumented Trail-Making Test. Gerontology 2022; 68: 234–240.

29.

Wiloth

, Lemke

, Werner

, et al. Validation of a computerized, game-based assessment strategy to measure training effects on motor-cognitive functions in people with dementia. JMIR Serious Games 2016; 4: e12.

30.

Mc Ardle

, Del Din

, Donaghy

, et al. The impact of environment on gait assessment: considerations from real-world gait analysis in dementia subtypes. Sensors 2021; 21: 813.

31.

Reis

, Postolache

, Teixeira

, et al. Exergames for motor rehabilitation in older adults: an umbrella review. Phys Ther Rev 2019; 24: 84–99.

32.

Yang

Y-H

, Lee

Y-H

, Yen

C-W

, et al. Association between cerebral coordination functions and clinical outcomes of Alzheimer’s dementia. Brain Sci 2022; 12: 1370.

33.

Suzumura

, Osawa

, Nagahama

, et al. Assessment of finger motor skills in individuals with mild cognitive impairment and patients with Alzheimer’s disease: Relationship between finger-to-thumb tapping and cognitive function. Jpn J Compr Rehabil Sci 2016; 7: 19–28.

34.

Suzumura

, Kanada

, Osawa

, et al. Assessment of finger motor function that reflects the severity of cognitive function. Fujita Med J 2021; 7: 122–129.

35.

Suzumura

, Osawa

, Kanada

, et al. Finger Tapping Test for assessing the risk of mild cognitive impairment. Hong Kong J Occup Ther 2022; 35: 137–145.

36.

Holmes

, Tripathi

, Katz

, et al. A novel framework to estimate cognitive impairment via finger interaction with digital devices. Brain Commun 2022; 4: fcac194.

37.

Yan

, Rountree

, Massman

, et al. Alzheimer’s disease and mild cognitive impairment deteriorate fine movement control. J Psychiatr Res 2008; 42: 1203–1212.

38.

Rovini

, Galperti

, Manera

, et al. A wearable ring-shaped inertial system to identify action planning impairments during reach-to-grasp sequences: a pilot study. J NeuroEngineering Rehabil 2021; 18: 118.

39.

Wang

, Wang

, Liu

, et al. Random forest model in the diagnosis of dementia patients with normal Mini-Mental State Examination scores. J Pers Med 2022; 12: 37.

40.

Garg

, Gupta

, Agarwal

, et al. Latest trends in outcome measures in dementia and mild cognitive impairment trials. Brain Sci 2022; 12: 922.