Replicated ECG-based Autonomic Profiles for Depression

Abstract Electrocardiography-derived autonomic profiles have been proposed as accessible markers of depression and suicide risk, but whether such a profile keeps its meaning outside the population in which it was derived has not been tested. We attempted a constrained reproduction of a published four-profile model in an independent, differently ascertained cohort: 35,980 hospital patients from MIMIC-IV-ECG, using 10-second routine clinical ECGs and ICD-10 diagnoses. Assigning patients to the published centroids without re-fitting, the heart-rate-variability magnitude axis reproduced: membership of a low-HRV profile was associated with depression (age- and sex-adjusted OR 1.19, 95% CI 1.13–1.26), anxiety (1.22, 1.15–1.29) and antidepressant prescription (1.32, 1.26–1.38), and the profile ranking held. The axis separating the two low-HRV profiles did not reproduce, consistent with an attenuation we quantified in advance from the spectral reconstruction the record length forced on us. Standardizing patients on the original cohort's statistics collapsed 75.1% of them into a single profile, showing that a centroid-based profile carries its meaning only together with the population frame in which it was computed. In a nonlinear extension over 123,932 patients, Higuchi fractal dimension of the raw 12-lead waveform was lower in depression (d = -0.39), anxiety (-0.42) and suicidality (-0.70), consistent with a transdiagnostic loss of physiological complexity rather than a depression-specific marker. In the same cohort, depression prevalence ranged from 26.2% to 51.3% depending only on which diagnostic codes were admitted. Autonomic profiles are reproducible in routine care data, but they are not portable without their coordinate system.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-13
DOI
https://doi.org/10.21203/rs.3.rs-11483537/v1
Primary Topic
Heart Rate Variability and Autonomic Control
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Replicated ECG-based Autonomic Profiles for Depression

Inkyung Choi, Milena Čukić, Yulim Choi
Research Square
Heart Rate Variability and Autonomic Control
preprint

Replicated ECG-based Autonomic Profiles for Depression

Inkyung Choi, Milena Čukić, Yulim Choi
preprint en

Abstract

Abstract Electrocardiography-derived autonomic profiles have been proposed as accessible markers of depression and suicide risk, but whether such a profile keeps its meaning outside the population in which it was derived has not been tested. We attempted a constrained reproduction of a published four-profile model in an independent, differently ascertained cohort: 35,980 hospital patients from MIMIC-IV-ECG, using 10-second routine clinical ECGs and ICD-10 diagnoses. Assigning patients to the published centroids without re-fitting, the heart-rate-variability magnitude axis reproduced: membership of a low-HRV profile was associated with depression (age- and sex-adjusted OR 1.19, 95% CI 1.13–1.26), anxiety (1.22, 1.15–1.29) and antidepressant prescription (1.32, 1.26–1.38), and the profile ranking held. The axis separating the two low-HRV profiles did not reproduce, consistent with an attenuation we quantified in advance from the spectral reconstruction the record length forced on us. Standardizing patients on the original cohort's statistics collapsed 75.1% of them into a single profile, showing that a centroid-based profile carries its meaning only together with the population frame in which it was computed. In a nonlinear extension over 123,932 patients, Higuchi fractal dimension of the raw 12-lead waveform was lower in depression (d = -0.39), anxiety (-0.42) and suicidality (-0.70), consistent with a transdiagnostic loss of physiological complexity rather than a depression-specific marker. In the same cohort, depression prevalence ranged from 26.2% to 51.3% depending only on which diagnostic codes were admitted. Autonomic profiles are reproducible in routine care data, but they are not portable without their coordinate system.

Research Square
ZHAW Zurich University of Applied Sciences (CH)
Good health and well-being
Heart Rate Variability and Autonomic Control
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