Deciphering the impact of chronological age on sleep stage transition dynamics for disorder classification

Ageing and sleep disorders are closely linked, as physiological changes can increase sleep fragmentation in older adults. Although sleep stage transition probability (STP) features provide useful markers of sleep disruption, age-related variability presents challenges for generalised sleep disorder classification. This study investigated the influence of chronological age on sleep disorder detectability and compared age-agnostic, age-involved, and age-specific machine learning (ML) approaches. Using single-night hypnograms across three data groups, two-stage STP features were extracted and evaluated through five experimental configurations, including models without age information, models incorporating age as a feature, and models stratified by age group. Statistical validation using absolute differences between transition probability matrices was performed to assess healthy-disordered differences across age cohorts and thresholds. Age-involved models provided modest improvements of approximately 2-4% in accuracy and F1-score over age-agnostic models, which achieved approximately 85% performance. Age-specific modelling demonstrated greater benefits, particularly among older subjects (age≥60), with improvements of approximately 8-12%. Older cohorts generally achieved 12-18% higher performance than younger cohorts, with XGBoost and other ensemble models producing the strongest results. In contrast, younger age-specific models showed lower or more variable performance, suggesting reduced discriminability of two-stage STP features. Statistical analysis identified more pronounced healthy-disordered differences in NREM transitions involving N2 and N3 among older subjects. Age-threshold analysis indicated that discriminability increased from the 50s and was strongest around 55-60 years. These findings suggest that chronological age influences the diagnostic relevance of hypnogram-derived transition features and that age-specific modelling may offer advantages for sleep disorder screening, particularly in older adults. However, independent external and prospective validation remains necessary before clinical implementation.

Authors

Institutions

Publication Details

Journal
PLOS Digital Health
Published
2026-11-16
DOI
https://doi.org/10.1371/journal.pdig.0001800
Primary Topic
Sleep and related disorders
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Deciphering the impact of chronological age on sleep stage transition dynamics for disorder classification

Fei He, Maia N Angelova, Ahsan Habib, Emran Ali et al.
2 links
PLOS Digital Health
Sleep and related disorders
article

Deciphering the impact of chronological age on sleep stage transition dynamics for disorder classification

Fei He, Maia N Angelova, Ahsan Habib, Emran Ali, Alireza Daneshkhah, Chandan Karmakar
article en

Abstract

Ageing and sleep disorders are closely linked, as physiological changes can increase sleep fragmentation in older adults. Although sleep stage transition probability (STP) features provide useful markers of sleep disruption, age-related variability presents challenges for generalised sleep disorder classification. This study investigated the influence of chronological age on sleep disorder detectability and compared age-agnostic, age-involved, and age-specific machine learning (ML) approaches. Using single-night hypnograms across three data groups, two-stage STP features were extracted and evaluated through five experimental configurations, including models without age information, models incorporating age as a feature, and models stratified by age group. Statistical validation using absolute differences between transition probability matrices was performed to assess healthy-disordered differences across age cohorts and thresholds. Age-involved models provided modest improvements of approximately 2-4% in accuracy and F1-score over age-agnostic models, which achieved approximately 85% performance. Age-specific modelling demonstrated greater benefits, particularly among older subjects (age≥60), with improvements of approximately 8-12%. Older cohorts generally achieved 12-18% higher performance than younger cohorts, with XGBoost and other ensemble models producing the strongest results. In contrast, younger age-specific models showed lower or more variable performance, suggesting reduced discriminability of two-stage STP features. Statistical analysis identified more pronounced healthy-disordered differences in NREM transitions involving N2 and N3 among older subjects. Age-threshold analysis indicated that discriminability increased from the 50s and was strongest around 55-60 years. These findings suggest that chronological age influences the diagnostic relevance of hypnogram-derived transition features and that age-specific modelling may offer advantages for sleep disorder screening, particularly in older adults. However, independent external and prospective validation remains necessary before clinical implementation.

PLOS Digital HealthVol. 5(11)
Deakin University (AU), Aston University (GB), Bulgarian Academy of Sciences (BG), Institute of Biophysics and Biomedical Engineering (BG), Emirates Aviation University (AE), Coventry University (GB)
Good health and well-being
Sleep and related disorders
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon