DEERFIELD BEACH, Fla., Aug. 13, 2026 (GLOBE NEWSWIRE) -- OTraces Inc. today announced results from its retrospective analysis of The Environmental Determinants of Diabetes in the Young (TEDDY) study data, in which its proprietary analytical methods identified children who later developed stage 3 type 1 diabetes (T1D) with overall predictive power above 92% at up to five years before clinical diagnosis.
The company's Proteomic Noise Suppression technology combined with Spatial Proximity classification achieved negative predictive power (NPV) of 92.9%, 94.9%, and 89.2% and positive predictive power (PPV)of 96.0%, 97.1% and 83.3% for 1-year, 3-year, and 5-year prediction horizons, respectively. In this analysis, those results exceeded both conventional logistic regression methods applied to the same data and previously published TEDDY-based models.
Model Performance Across Prediction Horizons
In the comprehensive analysis of 590 T1D-positive and 568 T1D-negative study participants from the multinational TEDDY cohort, OTraces models analyzed four biomarker concentrations: zinc transporter 8 (ZnT8) autoantibodies, mIAA, IA2A, and GADA.
When compared to conventional methods on identical biomarker sets, logistic regression achieved sensitivity of only 29.1%, 54.4%, and 40.0% for the same 1-, 3-, and 5-year prediction tasks. Published TEDDY-based autoantibody models showed specificities of just 35-50% at 90% sensitivity, compared to OTraces' 84-94% specificity range.
"The ability to identify children at risk for T1D years before onset represents a transformative opportunity for early intervention and clinical trial recruitment," said Alain Cappeluti, Co-founder and President, OTraces. "Our noise suppression methods capture the signature of disease progression by selectively compressing measured protein data to reveal patterns that conventional analytics miss."
Proprietary Technology Addresses Biological Complexity
OTraces' approach addresses a fundamental challenge in biological diagnostics: biomarker noise caused by biological complexity. The company's Proteomic Noise Suppression method partitions biomarkers into disease-coupled (noisy) and fixed variables (age, DNA markers, sex), then compresses disease-coupled concentration space into four zones defined by disease versus non-disease mean values.
This compression creates proximity scores that remove age-dependent drift and suppress variance not informative for T1D status, enabling multidimensional clustering that consistently produces 20-25 percentage points higher predictive power than logistic regression and 15-20 points improvement over classic neighborhood search or support vector machine methods.
Clinical and Research Implications
The predictive models offer multiple applications for managing T1D risk in high-risk populations. The ability to differentiate between 1-, 3-, and 5-year progression timeframes with high accuracy provides a powerful tool to forewarn at-risk individuals and recruit subjects for clinical trials exploring both disease triggers and therapeutic interventions. For sponsors developing disease-modifying therapies, stratifying candidates by likely time to onset could allow prevention trials to enroll participants most likely to progress within the study window, an approach that may reduce the number of participants required, shorten study duration, and strengthen statistical power.
Additional OTraces classifiers designed to refine estimated time-to-onset among already-positive samples achieved 94-95% sensitivities and 88-95% specificities for distinguishing 1-versus-5-year and 1-versus-3-year progression windows.
The analysis encompassed approximately 3.3 million data rows from the TEDDY database, collated with diagnosis information to construct models using samples drawn within specific time windows before diagnosis.
Path Forward
The testing system's simplicity—from blood draw to biomarker measurement to software analysis—enables deployment with minimal invasiveness and cost, allowing broader screening of at-risk populations.
These findings come from a retrospective analysis of existing TEDDY cohort data and have not been evaluated in a prospective clinical study. OTraces’ analytical methods described here are for research use only and have not been cleared or approved by the U.S. Food and Drug Administration for the prediction or diagnosis of type 1 diabetes. This study analyzed data obtained from the NIDDK Central Repository under the applicable data use agreement.
About OTraces Inc.
OTraces Inc., based in Deerfield Beach, Florida, is a biomedical diagnostics company specializing in proprietary analytical methods for disease prediction and diagnosis through advanced proteomic and other biomarker data analysis. For more information, visit https://otraces.com/.
Media Contact(s):
Alain D. Cappeluti
Co-Founder, and President & CFO OTraces, Inc.
acappeluti@otraces.com
