Predicting Early Functional Decline from Longitudinal Laboratory and Vital Sign Trajectories: A Large-Scale Study Using the All of Us Research Program

Rashmita Kudamala; Aravind V. Kuruvikkattil; Lalitha Pranathi Pulavarthy; Saptarshi Purkayastha

cs.LGAug 21, 2026

Predicting Early Functional Decline from Longitudinal Laboratory and Vital Sign Trajectories: A Large-Scale Study Using the All of Us Research Program

Authors: Rashmita Kudamala, Aravind V. Kuruvikkattil, Lalitha Pranathi Pulavarthy, Saptarshi Purkayastha

Organizations: Dept. of Biomedical Engineering and Informatics, Luddy School of Informatics, Computing, and Engineering, Indiana University Indianapolis, IN, USA

Abstract

Functional decline in older adults is typically recognized only after falls or observable gait impairment, closing the window for prevention. We investigated whether temporal trajectories of routine biomarkers, already recorded but rarely analyzed longitudinally, can identify patients in the pre-clinical phase of mobility decline. Using the All of Us Research Program (N = 297,861; 11.1% cases), we derived trajectory features (slope, variability, delta, mean) for twelve biomarkers over a three-year pre-index window. LightGBM models incorporating trajectories significantly outperformed static laboratory summaries (AUROC 0.797 vs. 0.755; DeLong p < 0.001; AUPRC 0.380 vs. 0.304). A 1:1 age- and sex-matched analysis confirmed an independent trajectory signal (AUROC 0.727 vs. demographics-only 0.680). A horizon analysis demonstrated sustained prediction 3-12 months before decline onset (AUROC 0.768-0.740). Because the model uses only measurements already ordered in routine care, it supports passive, zero-burden EHR integration for early detection of pre-clinical functional decline.

Figures & tables

Figure 1: Differential missingness by lab feature, stratified by outcome (3-year window). Cases show consistently lower missingness, consistent with informative observation in EHR data - patients developing functional impairment receive more frequent monitoring.

Feature Set	Model	AUC	95% CI	AUPRC
Demographics only	LGB	0.680	[0.677–0.683]	0.204
Baseline labs (3yr)	LR	0.640	–	–
	RF	0.714	–	–
	XGB	0.750	–	–
	LGB	0.755	[0.752–0.757]	0.304
Trajectory (3yr)	LR	0.726	–	–

Table 1: Model performance (3-year window, 5-fold CV, OOF; N = 297,861; 11.1% cases). Bootstrap 95% CI and AUPRC reported for LightGBM. LR = logistic regression; RF = random forest; XGB = XGBoost; LGB = LightGBM.

Figure 2: Precision–recall curves. The trajectory model (AP = 0.380) achieves 25% relative improvement over baseline (AP = 0.304). Dashed line: 11.1% prevalence.

Prediction Horizon	AUROC	AUPRC
3 months	0.768	0.361
6 months	0.756	0.358
9 months	0.748	0.355
12 months	0.740	0.353

Table 2: Prediction horizon performance using trajectory features. Measurements within the specified horizon prior to the index date were excluded.

Figure 3: SHAP beeswarm plot (held-out test set, $n$ = 5,000). Red: high feature values. BMI mean, blood pressure slopes, RDW mean, and systolic BP mean appear among the top physiologic predictors, with trajectory-specific features (slope, std) complementing static means.

Figure 4: Calibration curves. Isotonic regression substantially improves reliability (Brier 0.083 vs. 0.174) with AUC preserved.

Clinical Outcome Prediction

Clinical outcome prediction uses patient records, imaging, signals, and biomarkers to estimate risks, disease trajectories, treatment response, and survival.

See all 268 papers

Longitudinal Multimodal Sensing of Physical Activity and Well-Being in Older Adults
May 29, 2026Flavio Di Martino, Mattia G. Campana, Marcello Magno +2Clinical Outcome Prediction Wearable Sensing
IIT-CNR, Pisa, Italy · IFC-CNR, Pisa, Italy
A Temporal Machine Learning-Based Time-to-Event Model for Predicting ALS Progression and Healthcare Utilization
Jul 15, 2026Zongliang Yue, Qi Li, Terry Heiman-Patterson +3Event Time Prediction Survival Analysis
Department of Health Outcomes Research and Policy, Harrison College of Pharmacy, Auburn University Auburn, AL, USA
A Gait Foundation Model Predicts Multi-System Health Phenotypes from 3D Skeletal Motion
Mar 26, 2026Adam Gabet, Sarah Kohn, Guy Lutsker +11Gait Analysis Clinical Prediction
Department of Computer Science and Applied Mathematics, Weizmann Institute of Science, Rehovot, Israel. · Department of Molecular Cell Biology, Weizmann Institute of Science, Rehovot, Israel. · Newton VR LTD., Tel Aviv-Yafo, Israel +2

May 29, 2026J/K move · Enter open · S save

Longitudinal Multimodal Sensing of Physical Activity and Well-Being in Older Adults

Flavio Di Martino, Mattia G. Campana, Marcello Magno, Lorenza Pratali+1

IIT-CNR, Pisa, Italy · IFC-CNR, Pisa, Italy

Clinical Outcome Prediction Wearable Sensing Clinical Prediction Wearable Health Monitoring Multimodal Sensor Fusion

Open paper AskPDF arXiv

Wearable and mobile sensing technologies enable continuous monitoring of human behavior and health in real-world settings. However, predictive modeling in longitudinal multimodal data remains challenging, particularly when targeting complex or clinically derived outcomes. In this work, we present a longitudinal multimodal study of 66 older adults conducted in real-world conditions and combining wearable sensing, behavioral monitoring, and clinical assessments. This setting provides a rare opportunity to study an underrepresented population in long-term, into-the-wild conditions. Building on this dataset, we investigate how the alignment between sensed signals and target variables affects predictive performance across health-related tasks. We design a unified evaluation framework spanning tasks with increasing levels of observability, including Activity Levels prediction, Sleep Duration estimation, and Sleep Apnea Severity classification. Our results reveal a clear gradient of predictability: highly observable behavioral targets achieve robust performance (macro-F1 65%), while more abstract outcomes remain challenging despite consistent improvements over baseline models. Moreover, through explainability analysis, we show that historical features consistently emerge as the most informative predictors, highlighting the central role of longitudinal information.

Abstract

Clinical Outcome Prediction

Explore similar work

Longitudinal Multimodal Sensing of Physical Activity and Well-Being in Older Adults

A Temporal Machine Learning-Based Time-to-Event Model for Predicting ALS Progression and Healthcare Utilization

A Gait Foundation Model Predicts Multi-System Health Phenotypes from 3D Skeletal Motion