Peer-Reviewed Clinical Evidence

Clinical Evidence & Design Rationale

The scientific basis for every design decision: why frailty screening matters, why SPPB is the right instrument, why smartphone pose estimation is clinically valid, and why every biomechanical metric we compute is grounded in fall-risk research.

01

The Problem: Frailty and Falls in Ageing Singapore

6.2%
of Singaporeans 65+ are frail - most unaware until a fall occurs
37%
are pre-frail, placing them at elevated risk of rapid decline
684K
fatal falls globally per year; adults over 60 at highest risk

Frailty Prevalence in Singapore

Merchant RA, Chen MZ, Tan LWL, et al.
J Am Med Dir Assoc. 2017 - Singapore HOPE Study
Among community-dwelling Singaporeans aged 65+, 6.2% are frail and 37% are pre-frail - most unaware of their status until a fall or hospitalisation occurs.
PMID 28623152

What Frailty Predicts

Fried LP, Tangen CM, Walston J, et al.
J Gerontol A Biol Sci Med Sci. 2001;56(3):M146-56
Frailty - unintentional weight loss, exhaustion, weakness, slow walking speed, low physical activity - independently predicts falls, disability, hospitalisation, and mortality over 3 years, operating independently from comorbidity.
PMID 11253156

Falls Burden

World Health Organization
WHO Fact Sheet, 2024
Globally, 684,000 fatal falls occur annually. An additional 37.3 million falls require medical attention each year; 20-30% of older adults who fall suffer moderate to severe injuries including hip fractures and head trauma.
WHO Link

The Screening Gap

Clinical frailty screening requires trained professionals and in-person visits, limiting frequency. Singapore's push for AI-driven healthcare - including the NUS-Synapxe-IMDA AI Innovation Challenge 2026 - calls for tools that enable continuous remote monitoring and empower patients to manage health from home.

02

Why SPPB

The Short Physical Performance Battery (SPPB) assesses lower-extremity function through three timed tests: standing balance, gait speed, and repeated chair stands. Scored 0-12, it is the most widely validated geriatric mobility assessment.

Reason for Choosing SPPB Evidence
Predicts disability and mortality Guralnik et al. (1994) - 5,000+ subjects, strong predictive validity. PMID 8126356
Predicts disability onset Guralnik et al. (1995) - SPPB predicts disability over 4 years in persons 70+. NEJM
Predicts falls Lauretani et al. (2019) - SPPB ≤6 independently associated with falls; non-inferior to POMA. PMID 30515724
Quick and useful for stratification Treacy & Hassett (2020) - Effective risk stratification over 1- and 4-year follow-up. PMID 33191134
Observable via video Balance (sway), gait (cadence/symmetry), chair-stand (rep timing) - all have quantifiable biomechanical markers extractable from 2D pose estimation
03

Why Smartphone-Based Computer Vision Works

SilverGait uses MoveNet Lightning for real-time 17-keypoint 2D pose estimation on-device. Multiple studies validate that 2D markerless pose estimation produces clinically acceptable kinematics.

Can 2D Pose Estimation Replace Motion Capture?

Ung et al. (2022)
Gait & Posture - Direct MoveNet validation
MoveNet Thunder: 3.7 +/- 1.3 degrees mean hip angle error vs Vicon motion capture. Directly validates MoveNet for gait analysis.
PMID 35988434
Stenum et al. (2021)
PLoS Comput Biol
OpenPose: 4.0 degrees hip, 5.6 degrees knee, 7.4 degrees ankle error; 0.02s temporal error. Clinically acceptable for gait analysis.
PMC8099131
Bosquet et al. (2023)
Front Rehabil Sci - Elderly-specific
2D pose estimation comparable to marker-based systems specifically for elderly gait analysis in clinical settings.
View Paper
Ali et al. (2024)
Evol Bioinform - MediaPipe validation
MediaPipe: Pearson's r = 0.80 lower limb, 0.91 upper limb vs Qualisys gold-standard motion capture.
View Paper

Can a Smartphone Do This?

Shin et al. (2021)
Front Physiol
Smartphone at 30fps produces clinically usable gait analysis. Validates the feasibility of mobile-first pose estimation for gait.
View Paper
Gong et al. (2025)
Smart Health
Validates smartphone-based joint angle analysis for sit-to-stand in elderly - directly supporting our chair-stand test implementation.
View Paper
Roldan-Jimenez et al. (2025)
Gait & Posture
Mobile phone pose estimation: ICC >0.9 for sit-to-stand - excellent reliability for clinical use.
View Paper
Patel et al. (2023)
PLOS Digit Health
Video-based pose estimation can detect clinically meaningful change over time, supporting longitudinal monitoring use cases.
View Paper

Systematic Reviews

Scott et al. (2023)
J NeuroEng Rehabil - Systematic Review
Markerless motion capture strong for postural control and gross movement across diverse clinical populations.
View Paper
Ripic et al. (2024)
Sensors - Systematic Review
Markerless camera-based capture reliable and valid for hip/knee gait kinematics, supporting clinical deployment of camera-only systems.
View Paper
04

Why We Chose Each Metric

Every biomechanical metric SilverGait extracts is grounded in clinical fall-risk research. Here's why we measure what we measure.

Joint Angles - Three-Point Dot-Product Formula
angle_at_B = arccos( (BA . BC) / (|BA| x |BC|) )

This formula is view-invariant in the sagittal plane - unlike raw pixel coordinates, joint angles don't change with camera distance or position.

  • Why knee angle: Knee flexion/extension range is a key descriptor of sit-to-stand ability and gait phase. Roebroeck et al. (2007) identify knee extension velocity as a primary STS descriptor. Link
  • Why hip angle: Trunk-to-thigh flexion during STS is a compensatory movement indicator. Excessive forward lean signals lower-limb weakness.
  • Confidence threshold (0.3): Intentionally low - MoveNet Lightning prioritizes speed, and temporal aggregation smooths noisy readings.
Sway Velocity & Area (Hip Center Displacement)

We track the midpoint of left/right hip keypoints as a center-of-mass proxy and compute frame-to-frame displacement (sway velocity) and bounding-box area (sway area).

75-90% increased fall risk in the highest postural sway quintile (Melzer et al. 2019)

We use hip center - not nose or shoulders - because it's the most stable CoM proxy from 2D keypoints. Nose is affected by head turns; shoulder midpoint by arm movement.

Melzer 2019 - PMID 30981147
Trunk Lean Variability (SD of Trunk Angle)

We compute trunk lean as the angle between the shoulder-hip line and vertical, then track its standard deviation over time.

Gill et al. (2001) - elderly show significantly greater trunk sway during clinical balance tests. Trunk lean variability captures postural instability that mean values miss. Link

Gait Rhythm Variability (CV of Inter-Step Intervals)

We detect steps via zero-crossings of smoothed ankle X separation, then compute the coefficient of variation of inter-step intervals.

Stride time variability was 106ms in fallers vs 49ms in non-fallers - the strongest gait-based fall predictor in a 1-year prospective study (Hausdorff et al. 2001)
Hausdorff 2001 - PMID 11494184
Step Symmetry Index
abs(avgLeft - avgRight) / avgBoth x 100%

Gait asymmetry combined with cadence and double stance time predicts falls with AUC 0.845 (Chen et al. 2024). PMC11323353

Cadence & Normative Ranges
cadence = (stepCount / durationSec) x 60

Hollman et al. (2011) normative spatiotemporal gait parameters in older adults provide healthy/at-risk thresholds. PMC3104090

Chair-Stand Rep Time Consistency (CV of Rep Durations)

We detect reps via prominence-based valley detection in the hip Y signal, then compute the CV of rep-to-rep durations.

  • Rep timing reliability: Bohannon (2021) - high ICC for 5xSTS timing validates rep time consistency as a reliable, reproducible measure. PMC8228261
  • Prominence-based detection: Fixed Y-thresholds fail because camera distance, chair height, and user height change absolute values. Prominence detection is self-calibrating.

Derived Metrics Over Raw Coordinates

We compute ~16 clinically meaningful features per frame and discard raw coordinates immediately. This is by design:

Problem with Raw Keypoints How Derived Metrics Solve It
Camera distance changes pixel values Joint angles are view-invariant in the sagittal plane
MoveNet jitters several pixels frame-to-frame Temporal aggregation (mean, CV) filters noise
15,300 values per 20s recording bloat LLM prompt ~30 scalar summary leaves context for Gemini's video analysis
05

Why Sleep Intervention

SilverGait includes a Sleep Agent that generates personalized CBT-I and sleep hygiene plans based on each user's frailty tier, mood risk, exercise streak, and social isolation level.

Sleep and Frailty Are Bidirectional

Poor sleep accelerates muscle loss and frailty progression, while frailty itself worsens sleep quality - creating a vicious cycle.

Ensrud et al. (2012)
J Am Geriatr Soc. - 3,000+ older women, 7-year study
Poor sleep quality associated with increased frailty risk in 3,000+ older women over 7 years of follow-up.
PMID 22283806
Moreno-Tamayo et al. (2020)
Sleep Med.
Insomnia symptoms predict frailty incidence in community-dwelling elderly. Bidirectional relationship confirmed - each condition worsens the other.
PMID 32298918

CBT-I Is the Gold Standard for Elderly Insomnia

Pharmacological sleep aids carry fall risks for elderly. CBT-I is recommended as first-line treatment.

Irwin et al. (2006)
J Am Geriatr Soc.
CBT-I produces durable improvements in sleep quality in older adults, without the fall risks associated with pharmacological options.
PMID 16551307
Sivertsen et al. (2006)
JAMA - Head-to-head comparison
CBT-I superior to zopiclone (sleep medication) at 6-month follow-up in older adults - making it the evidence-based first-line choice.
PMID 16790700
The Sleep Agent personalizes advice based on contextual factors: mood risk (anxiety worsens sleep), exercise streak (physical activity improves sleep), frailty tier (poor sleep accelerates muscle loss), and social isolation (loneliness linked to poor sleep in elderly). For moderate/high risk users, CBT-I techniques (sleep restriction, stimulus control, progressive muscle relaxation) are included.
06

Why Exercise Tracking & Personalization

Exercise Prevents and Reverses Frailty

Cadore et al. (2013)
Ageing Res Rev.
Multicomponent exercise (strength + balance + gait) is the most effective intervention for frailty reversal in elderly populations.
PMID 23266332
de Labra et al. (2015)
BMC Geriatr. - Systematic review of 19 RCTs
Exercise programs improve physical function, mobility, and balance in frail older adults across 19 randomized controlled trials.
PMID 26126532

Exercise Reduces Fall Risk

Sherrington et al. (2019)
Br J Sports Med. - 108 RCTs, Cochrane-level
Exercise reduces fall rate by 23% overall. Programs including balance training reduce falls by 39%. The most definitive evidence base in the field.
PMID 30464052
Gillespie et al. (2012)
Cochrane Database Syst Rev.
Multifactorial interventions including exercise reduce rate of falls in community-dwelling older adults - supporting our integrated approach.
PMID 22972103

Our Approach: Safety-Tiered Intensity

Exercise plans are selected deterministically from a curated content library by frailty tier - not LLM-generated - ensuring safety-appropriate intensity.

Robust
20-30 min
Moderate intensity maintenance
Pre-frail
15-20 min
Strengthening for balance & leg power
Frail
10-15 min
Gentle seated/supported exercises
Severely Frail
5-10 min
Minimal caregiver-assisted movements

The Exercise Agent (LLM) provides additional personalization based on specific SPPB deficits (low balance, slow gait, weak chair stand) and exercise streak - layered on top of the safe deterministic base.

07

Evidence Strength Summary

A consolidated view of the clinical evidence underpinning each component of SilverGait.

Component Evidence Strength Key Papers
SPPB as clinical tool
Very Strong
Guralnik 1994, 1995
SPPB predicts falls
Strong
Lauretani 2019, Treacy 2020
Frailty predicts adverse outcomes
Very Strong
Fried 2001
Singapore frailty prevalence
Strong
Merchant 2017 (HOPE Study)
MoveNet for 2D kinematics
Strong
Ung 2022
Smartphone-based gait analysis
Strong
Shin 2021
Smartphone-based STS analysis
Strong
Gong 2025, Roldan-Jimenez 2025
Gait variability as fall predictor
Very Strong
Hausdorff 2001
Postural sway as fall predictor
Strong
Melzer 2019
Trunk lean variability
Strong
Gill 2001
Step symmetry as fall predictor
Moderate-Strong
Chen 2024
STS rep time consistency
Strong
Bohannon 2021
Markerless motion capture validity
Strong
Scott 2023, Ripic 2024
Sleep-frailty bidirectional link
Strong
Ensrud 2012, Moreno-Tamayo 2020
CBT-I for elderly insomnia
Very Strong
Irwin 2006, Sivertsen 2006
Exercise reverses frailty
Very Strong
Cadore 2013, de Labra 2015
Exercise reduces falls
Very Strong
Sherrington 2019 (108 RCTs)