Insulin Resistance Prediction From Wearables and Routine Blood Biomarkers

Ahmed A. Metwally, A. Ali Heydari, Daniel McDuff, Alexandru Solot, Zeinab Esmaeilpour, Anthony Z. Faranesh, Menglian Zhou, David B. Savage, Conor Heneghan, Shwetak Patel, Cathy Speed, Javier L. Prieto · 2025

Read Insulin Resistance Prediction From Wearables and Routine Blood Biomarkers by Ahmed A. Metwally, A. Ali Heydari, Daniel McDuff, Alexandru Solot, Zeinab Esmaeilpour, Anthony Z. Faranesh, Menglian Zhou, David B. Savage, Conor Heneghan, Shwetak Patel, Cathy Speed, Javier L. Prieto (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-03784.txt.

Introduction

Metwally and co-authors predict insulin resistance from wearable data and routine blood tests, in a research paper from 2025. Insulin resistance paves the road to type 2 diabetes, yet the standard lab measure is costly and hard to access. Early flags could open the door to prevention.

The team recruited 1,165 people across the United States, the largest such set to date, with wearable time series and blood markers. A deep network that fused both sources beat either source alone at forecasting the standard resistance score. It did best in obese and sedentary users, the group at highest risk. A fresh cohort of 72 people confirmed the scores. The team also fed the forecast into a language model agent for safe, plain guidance.

Catalog record

CREATORAhmed A. Metwally, A. Ali Heydari, Daniel McDuff, Alexandru Solot, Zeinab Esmaeilpour, Anthony Z. Faranesh, Menglian Zhou, David B. Savage, Conor Heneghan, Shwetak Patel, Cathy Speed, Javier L. Prieto
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE102 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-03784

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Insulin Resistance Prediction From Wearables and Routine Blood Biomarkers Ahmed A. Metwally, A. Ali Heydari, Daniel McDuff, Alexandru Solot, Zeinab Esmaeilpour, Anthony Z. Faranesh, Menglian Zhou, David B. Savage, Conor Heneghan, Shwetak Patel, Cathy Speed, Javier L. Prieto Abstract Insulin resistance, a precursor to type 2 diabetes, is characterized by impaired insulin action in tissues. Current methods for measuring insulin resistance, while effective, are expensive, inaccessible, not widely available and hinder opportunities for early intervention. In this study, we remotely recruited the largest and most geographically, age and gender diverse dataset to date across the US to study insulin resistance (N=1,165 participants, with median BMI=28 kg/m 2 , age=45 years, HbA1c=5.4%), incorporating wearable device time series data and blood biomarkers, including the ground-truth measure of insulin resistance, homeostatic model assessment for insulin resistance (HOMA-IR). We developed deep neural network models to predict insulin resistance based on readily available digital and blood biomarkers. Our results show that our models can predict insulin resistance by combining both wearable data and readily available blood biomarkers better than either of the two data sources separately (R 2 =0.5, auROC=0.80, Sensitivity=76%, and specificity 84%, based on HOMA-IR cut-off of 2.9, using wearables, demographics, and routine blood biomarkers). The model showed 93% sensitivity and 95% adjusted specificity in obese and sedentary participants, a subpopulation most vulnerable to developing type 2 diabetes and who could benefit most from early intervention. Adjusted specificity emphasizes false positives (predicting insulin resistance in truly insulin-sensitive individuals and not impaired insulin sensitivity). Rigorous evaluation of model performance, including interpretability, robustness, and sensitivity, facilitates generalizability across larger cohorts, which is demonstrated by reproducing the prediction performance on an independent validation cohort (N=72 participants). Additionally, we demonstrated how the predicted insulin resistance can be integrated into a large language model agent to help understand and contextualize HOMA-IR values, facilitating interpretation and safe personalized recommendations. This work offers the potential for early detection of people at risk of type 2 diabetes and thereby facilitate earlier implementation of preventative strategies. Early identification could allow for lifestyle interventions that could reverse insulin resistance in those at highest risk of developing type 2 diabetes and could have a particularly profound impact among those who are unknowingly at risk. \reportnumber 1 Introduction Currently, 537 million adults are living with diabetes, a figure that is estimated to increase to 643 million by 2030, where approximately 10% of people with diabetes have Type 1 diabetes (T1D), while around 90% have Type 2 diabetes (T2D) ( 1-n 2022 ) . This rise of T2D will be driven largely by lifestyle behaviors ( Galaviz et al. 2018 ) . In a healthy individual, insulin, a hormone secreted by pancreatic beta-cell, helps regulate blood glucose levels by facilitating glucose uptake into cells (including muscles, adipose, and liver) from the blood. Additionally, incretin hormones, such as Glucagon-like Peptide-1 (GLP-1) and Gastric Inhibitory Polypeptide (GIP), can increase insulin secretion from pancreatic beta-cells, leading to improved glycemic control ( El et al. 2021 ) . The fundamental problem in diabetes is the inability of the body to regulate blood glucose properly due to absolute or relative insulin deficiency. In T1D, the body’s immune system mistakenly attacks and destroys the pancreatic beta-cells, resulting in absolute insulin deficiency and elevated blood glucose ( Roep et al. 2020 ) . Alternatively, in the majority of T2D, the body becomes insulin resistant, requiring higher amounts of insulin to be produced by the pancreatic beta-cells to achieve the same glucose-lowering effect. With time, the pancreatic beta cells might become unable to produce enough insulin to compensate for insulin resistance (IR), leading to relative insulin deficiency and elevated blood glucose levels. The prevalence of IR in the general population is estimated to be between 20% and 40%, with variations observed across different ethnic groups, age brackets, lifestyle, and the presence of comorbidities ( Parcha et al. 2021 ) . Chronic insulin resistance puts the person at significant risk for prediabetes and overt type 2 diabetes and is also strongly associated with metabolic dysfunction-associated steatotic liver disease (MASLD) and cardiovascular disease (CVD) risk. IR prevalence in type 2 diabetes is 83.9% ( Bonora et al. 1998 ) . Contributing factors include excess body weight, particularly visceral fat, physical inactivity, and genetic predisposition. Figure 1 A illustrates the complex nature of T2D and the intricate relationships between lifestyle choices, genetics, and various metabolic subphenotypes and physiological processes involved in the development of the disease. Insulin resistance may manifest subtly with high blood pressure, or abnormal lipid profiles. Long term complications of diabetes include damage to various organs and tissues over time, such as diabetic retinopathy, nephropathy, and neuropathy ( Sapra and Bhandari 2022 ) . Moreover, insulin resistance is part of the pathophysiology of cardiovascular-kidney-metabolic (CKM) syndrome, characterized by interactions among metabolic risk factors, chronic kidney disease, and the cardiovascular system. Insulin resistance is a key component of metabolic syndrome, which contributes to the development of cardiovascular disease ( Khan et al. 2023 ; Ndumele et al. 2023 ; Iglesies-Grau et al. 2023 ) . Early detection of insulin resistance and intervention could play a critical role in reducing CKM disease burden. Identifying IR early can guide several focused lifestyle interventions, like weight loss, regular exercise, and a healthy dietary pattern, that can substantially improve or even reverse insulin resistance. While most individuals can certainly benefit from most types of physical activity and healthy diets, there are specific interventions that have been scientifically proven to prevent and treat IR. For lifestyle interventions, aerobic training ( Turcotte and Fisher 2008 ) , resistance training ( Turcotte and Fisher 2008 ; Niemann et al. 2020 ) , calorie-restricted diets ( Zhang et al. 2021 ) , and low-fat diets ( Demaria et al. 2023 ) have all shown to be valuable in reducing IR. On the therapeutic end, Thiazolidinediones and Metformin medications have frequently been shown to reduce IR ( Ko et al. 2017 ; Sui et al. 2019 ) . Recent studies have shown that incretin hormone agonists, like GLP-1 and GIP, act as sensitizers and improve insulin resistance Garvey et al. 2022 ; Frias et al. 2024 . Several methods are available for measuring insulin resistance (IR), but are not implemented routinely, meaning that opportunities for early intervention are often missed. Instead, a focus on snapshots of glucose levels, fasting glucose, Hemoglobin A1c (HbA1c), or glucose level after a 2-hour oral glucose tolerance test (OGTT) represent the typical screening approach, and can be insensitive to those in early stages of IR. The gold standard test for IR is the hyperinsulinemic euglycemic clamp ( DeFronzo et al. 1979 ) , which is performed in research facilities only, expensive, and time-consuming. Homeostatic Model Assessment of Insulin Resistance (HOMA-IR) and minimal model-based glucose and insulin measures are affordable and faster alternatives that can be performed in clinical labs ( Matthews et al. 1985 ) . However, they require a clinical lab visit and may not be as accurate as the clamp. HOMA-IR is a mathematical model used to estimate the degree of insulin resistance in an individual ( Matthews et al. 1985 ) . It provides a simple way to assess how well the body is using insulin.

Read the complete text