The prediction of the ECG age led by AI transforms the detection of early diseases

A simple ECG scan could now predict your risk of heart disease, Alzheimer’s and cancer before symptoms appear – thanks to the monitoring of the biological age powered by AI.

Study: Reclassification of conventional risk assessment for diseases linked to organic aging compatible with electrocardiogram. Image credit: Totojang1977 / Shutterstock

In a recent study published in the journal NPJ agingThe researchers assessed whether the biological age estimated by the electrocardiogram compatible with artificial intelligence (AI) (ECG-BA) improves the classification of risks for diseases linked to aging beyond chronological age (CA) .

Background

Did you know that two people of the same age can have radically different health results? Aging affects individuals differently, some remain active and without illness while others develop serious conditions.

Aging is a universal process that leads to a physiological decline, increasing the risk of neurodegenerative, cardiovascular (CV), metabolic, musculoskeletal and immune. The turnover is commonly used in disease prediction models, but it does not capture the variability of biological aging between individuals. The study excluded people with pre -existing conditions such as hypertension, diabetes and cardiac incapacity to focus on a “healthy” population.

The ECG-BA, derived from physiological biomarkers, provides a more personalized measure of the state of health. AI now allows real-time analysis of ECG signals to estimate the ECG-BA, improving the stratification of risks. Crossed validation five times has been applied to optimize model performance, guaranteeing robust results. Additional research is necessary to validate its predictive value for various populations.

About the study

The study used ECG recordings collected from the general Taipei veterans hospital between 2006 and 2017. Initially, 51,061 ECG valids were recorded, but after applying exclusion criteria, 48,783 people in good health aged 20 to 80 were analyzed.

An in-depth learning model incorporating a residual network (RESNET), a compression and excitation network (SENET) and multitasking learning has been developed to estimate ECG-BA from 12 derivations. The model was optimized using the Adam optimizer, which refined network weights for improved precision. The CA and the medical records were linked to the help of international codes for the classification of diseases (ICD) to classify participants in groups of diseases and control linked to aging.

The formation of the model involved a five -year cross validation to optimize performance. The main assessment metric was the correlation between ECG-BA and CA in a healthy population.

The diagnostic performance for CV and non -CV diseases have been evaluated using areas in the receptor operating characteristics (ROC). Improvement of net reclassification (NRI) was calculated to measure the improvement of the risk classification after incorporating ECG-BA.

Statistical analysis included conditional logistical regression to assess the predictive usefulness of the model in the classification of diseases. The average absolute error of the model (MAE) was 6.25 years, with an absolute percentage error (MAP) of 15.35%, indicating a high predictive precision compared to the previous models.

The processing of data and the implementation of the model were carried out using Pytorch, with results validated in relation to the established clinical benchmarks.

Study results

Imagine being able to predict future health risks with a simple ECG, as is the way a smartwatch monitors daily heart activity. This study reveals that ECG-BA is a powerful tool for identifying diseases related to aging earlier and more precisely than AC alone.

The model showed a strong correlation between ECG-BA and CA (R² = 0.70, p <0.01). The predictive precision of the model was higher than the ECG models based on the previous AI, which had greater margins of error. However, the real value of this technology is its ability to identify people at risk of developing serious illnesses before the appearance of traditional symptoms.

Compared to the use solely of AC, the incorporation of the ECG-BA has considerably improved the prediction of the risk for conditions such as coronary disease (CAD), stroke and myocardial infarction (IM ).

For example, improving net reclassification (NRI) for peripheral arterial disease (PAOD) was 1.1% (from 0.8632 to 0.8653, p <0.01), which means that the ECG-BA refined the classification of risk beyond AC alone. The classification of cancer risks has improved by 29% in terms of NRI, demonstrating that this technology can refine medical assessments and more effectively target high -risk individuals.

For a real impact, consider cancer detection. Early diagnosis Maybe the difference between life and death. The study has shown that the ECG-BA has corrected 21% of classification errors made by AC alone, reducing the number of patients incorrectly classified. This means that more high -risk people could be identified earlier, which potentially allows timely interventions that save lives.

The most significant improvements have been observed in people aged 40 and over, strengthening the idea that biological aging – not only the number of years experienced – should be taken into account in health care assessments.

Despite its success in refining disease prediction, the model had limits to forecasting the conditions related to arrhythmia such as atrial fibrillation (AF) and sick sinus syndrome (SSS). The study suggests that arrhythmias are influenced by factors beyond aging, such as hyperthyroidism, smoking and lifestyle, which can explain why ECG-BA is less effective for these conditions.

However, for conditions motivated by biological aging, such as Alzheimer’s disease (AD) and osteoarthritis (OA), this tool has a revolutionary opportunity to improve early detection and preventive health strategies.

With the increase in accessibility to monitoring ECG via portable devices, these results have large -scale implications. However, the study notes that the ECG-BA models require additional validation on different ECG machines, such as Philips and Ge Healthcare, because variations in devices could have an impact on predictions.

Imagine a future where routine ECGs not only detect immediate heart problems, but also provide a personalized aging risk score, helping individuals take proactive measures to maintain long -term health.

This study marks an important step towards this future, demonstrating that the ECG-BA can reshape preventive medicine and risk assessment, ultimately improving the results for health on a global scale.

Conclusions

To summarize, the ECG-BA provides additional value in the risk classification for diseases related to aging beyond CA. The in -depth learning model has demonstrated significant improvements in predictive precision, in particular for CV conditions, MA, osteoarthritis and cancers.

Analysis of improving net reclassification (NRI) indicated that the integration of the ECG-BA could correct classification errors in 21% of cases, with the highest improvement (29%) observed in the prediction of the risk of cancer. The results highlight the potential of ECGs as a non -invasive and profitable biomarker for systemic aging.

However, the study also emphasizes the need for multicenter validation to confirm the generalization of various populations and platforms of devices.