AccelHealthDigital Phenome Atlas

AI-Powered AccelerometerHealth Hub

An integrated platform linking accelerometer-derived phenotypes (ADPs) to human health outcomes.

Discover associationsPredict disease riskExplore ADP trajectoriesIdentify exposure-response relationships
XYZ7 dayscontinuous monitoring3-axisacceleration dataXYZwearable device
N=70,473~9 yrs2013-2015Nov 2023large prospective cohort
ATSTime ResNet Module...ADPsFully Connected NetworkWeightedFusionAge+SexFCNNDiagnosisPredictionWATSNet
AI-Powered Analysis
TimeRiskHigh RiskLow Risk
Risk Prediction
70,473Participants
210ADPs
406Diseases
956Traits
Axivity AX3Accelerometer
N = 70,473
Healthy
Cases
Acceleration Time Series (ATS)
D1D2D3D4D5D6D7
Diseases406
Health Traits956
Accelerometer-derived Phenotypes (ADPs)
N = 210
Periods
...
Activities
AccelerationLIPAMVPASedentarySleepCircadian rhythmStepCadence
Statistics
MeanStdMedianMinMaxSkewnessKurtosisDuration
Association · 210 ADPs · 406 Diseases/956 Traits

Discover ADP-disease/trait
relationships

Discover associations between ADPs with full spectrum of diseases and health-related traits.

→
Association Manhattan Plot
Prediction & Diagnosis · WATSNet

AI powered disease prediction
and diagnosis.

Advanced AI integrates ADPs with ATS to enhance disease diagnosis and prediction.

→
KM Survival Curves
AUC Curves
Trajectories · 210 ADPs

Explore ADP dynamics during
pre-diagnostic periods.

ADPs exhibited dynamic changes as the time of disease diagnosis approached.

→
Feature Trajectories
Exposure-Response

Provide evidence-informed
behavioral benchmarks.

Quantifies how activity patterns may be associated with lower disease risks and provides benchmark levels.

→
Exposure-Response Curves