Exercise Stress Test Lactate Curves and Logistic Regression for Type 2 Diabetes Mellitus Detection: A Quartile-Based Analysis

Objective: This study aimed to evaluate the utility of exercise stress test lactate curves combined with logistic regression modeling for the detection of type 2 diabetes mellitus (T2DM) using quartile-based stratification of maximum speed at the lactate threshold (4 mmol/L). Methods: A cross-sectional study was conducted comparing T2DM patients and healthy controls. Lactate levels were measured in participants undergoing standardized exercise stress testing, with the lactate threshold set at 4 mmol/L. Subjects were stratified into quartiles (Q1–Q4) based on their maximum speed at the lactate threshold. Analysis of variance (ANOVA) was performed across 1428 tests examining physiological, biochemical, cognitive, and biomarker variables. Significant predictors identified through variable-selection procedures were incorporated into a logistic regression model. Results: ANOVA revealed 126 significant differences (8.8% overall) between T2DM patients and controls across all quartiles, with Q4 demonstrating the highest significance rate (12.8%, 46/358 tests). Key discriminatory variables included HbA1c (p < 0.01 in Q2, Q3, Q4), fasting glucose (p < 0.05 in Q2, Q3, Q4), systolic blood pressure (DER Systole, p < 0.05 across multiple quartiles), homocysteine (p < 0.01 in Q4), and brain-derived neurotrophic factor (BDNF, p < 0.05 in Q4). The logistic regression model, developed on a sample with 68.3% diabetes prevalence, identified three significant predictors: DER Systole (negative association), vitamin D (VIT_D, positive association), and lactate velocity (Vel_Lactato2, positive association). The model achieved moderate performance, with 84.5% accuracy, 97.9% sensitivity, 58.3% specificity, 82.1% precision, 93.3% negative predictive value, an F1-score of 0.882, an area under the ROC curve (AUC) of 0.792, and a Gini coefficient of 0.583. Conclusions: The integration of exercise stress test lactate curves with logistic regression modeling provides an adequate, minimally invasive tool for T2DM detection. The quartile-based analysis reveals that physiological and biochemical differences between diabetic and non-diabetic individuals are more pronounced among those with higher aerobic capacity (Q4). The predictive model demonstrates adequate discriminatory power (AUC = 0.792) and represents an exploratory hypothesis-generating approach for diabetes risk stratification in clinical and sports medicine settings.

Authors

Institutions

Publication Details

Journal
Diagnostics
Published
2026-11-18
DOI
https://doi.org/10.3390/diagnostics16223842
Primary Topic
Diabetes Management and Research
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Exercise Stress Test Lactate Curves and Logistic Regression for Type 2 Diabetes Mellitus Detection: A Quartile-Based Analysis

Martha Cecilia Jimenez Martinez, César Baracaldo, Sandra Rojas Vega
1 links
Diagnostics
Diabetes Management and Research
article

Exercise Stress Test Lactate Curves and Logistic Regression for Type 2 Diabetes Mellitus Detection: A Quartile-Based Analysis

Martha Cecilia Jimenez Martinez, César Baracaldo, Sandra Rojas Vega
article en

Abstract

Objective: This study aimed to evaluate the utility of exercise stress test lactate curves combined with logistic regression modeling for the detection of type 2 diabetes mellitus (T2DM) using quartile-based stratification of maximum speed at the lactate threshold (4 mmol/L). Methods: A cross-sectional study was conducted comparing T2DM patients and healthy controls. Lactate levels were measured in participants undergoing standardized exercise stress testing, with the lactate threshold set at 4 mmol/L. Subjects were stratified into quartiles (Q1–Q4) based on their maximum speed at the lactate threshold. Analysis of variance (ANOVA) was performed across 1428 tests examining physiological, biochemical, cognitive, and biomarker variables. Significant predictors identified through variable-selection procedures were incorporated into a logistic regression model. Results: ANOVA revealed 126 significant differences (8.8% overall) between T2DM patients and controls across all quartiles, with Q4 demonstrating the highest significance rate (12.8%, 46/358 tests). Key discriminatory variables included HbA1c (p < 0.01 in Q2, Q3, Q4), fasting glucose (p < 0.05 in Q2, Q3, Q4), systolic blood pressure (DER Systole, p < 0.05 across multiple quartiles), homocysteine (p < 0.01 in Q4), and brain-derived neurotrophic factor (BDNF, p < 0.05 in Q4). The logistic regression model, developed on a sample with 68.3% diabetes prevalence, identified three significant predictors: DER Systole (negative association), vitamin D (VIT_D, positive association), and lactate velocity (Vel_Lactato2, positive association). The model achieved moderate performance, with 84.5% accuracy, 97.9% sensitivity, 58.3% specificity, 82.1% precision, 93.3% negative predictive value, an F1-score of 0.882, an area under the ROC curve (AUC) of 0.792, and a Gini coefficient of 0.583. Conclusions: The integration of exercise stress test lactate curves with logistic regression modeling provides an adequate, minimally invasive tool for T2DM detection. The quartile-based analysis reveals that physiological and biochemical differences between diabetic and non-diabetic individuals are more pronounced among those with higher aerobic capacity (Q4). The predictive model demonstrates adequate discriminatory power (AUC = 0.792) and represents an exploratory hypothesis-generating approach for diabetes risk stratification in clinical and sports medicine settings.

DiagnosticsVol. 16(22)
Pedagogical and Technological University of Colombia (CO)
Universidad Pedagógica y Tecnológica de Colombia, Bundesministerium für Bildung und Forschung
Good health and well-being
Diabetes Management and Research
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