Periodic Reporting for period 1 - AI-POD (Trustworthy AI Tools for the Prediction of Obesity Related Vascular Diseases)
Reporting period: 2023-05-01 to 2024-10-31
Almost 60% of the entire European population have been diagnosed overweight or obese. Obesity causes more than 1.2 million deaths annually, corresponding to more than 13% of total mortality in the WHO European Region. Obesity is linked with clusters of diseases that greatly increase risk of cardiovascular disease (CVD), which is the most common cause of death in Europe (3.9 million deaths/year). As a consequence, those living with obesity are at substantially higher risk for a range of CVDs.
To date, predicting how CVD will progress in people with obesity remains a challenge. It has been recognised that obesity affects individuals differently. People with similar body mass indexes (BMIs) may have very different risks for cardiovascular and metabolic problems. Common clinical and imaging-based methods to assess risk (Framingham Risk Score and Agatston Score) as well as BMI do not reliably account for the complexity of an accurate evaluation regarding the development of obesity-related cardiac diseases. Important influencing factors of obesity on like plaque composition, plaque progression, or muscle loss, are not adequately considered by the current risk scores and parameters. Current standard risk prediction scores and methods do not include real-time monitoring of diet and lifestyle choices.
The main goal of the project ‘Trustworthy AI Tools for the Prediction of Obesity-Related Vascular Diseases (AI-POD)’ is to develop and validate AI tools for the assessment and prediction of risk of cardiovascular diseases and related complications in obese persons. AI-POD will advance the current state-of-the-art by introducing a novel, ground-breaking cardiovascular risk prediction and management approach for the obese population.
For the first time, all these factors will be considered in a novel, fair and trustworthy AI-based preventive risk score, integrating the influence of diet and lifestyle in real-time monitoring. AI-POD’s main outcomes are (1) a novel AI-based Clinical Decision Support System (CDSS) for the risk assessment (AI-POD risk score) and prediction of obesity-related CVDs and associated complications; (2) an innovative, easy-to-use mobile app for citizens (Citizen App) that interacts with the CDSS and considers lifestyle information.
To date, predicting how CVD will progress in people with obesity remains a challenge. It has been recognised that obesity affects individuals differently. People with similar body mass indexes (BMIs) may have very different risks for cardiovascular and metabolic problems. Common clinical and imaging-based methods to assess risk (Framingham Risk Score and Agatston Score) as well as BMI do not reliably account for the complexity of an accurate evaluation regarding the development of obesity-related cardiac diseases. Important influencing factors of obesity on like plaque composition, plaque progression, or muscle loss, are not adequately considered by the current risk scores and parameters. Current standard risk prediction scores and methods do not include real-time monitoring of diet and lifestyle choices.
The main goal of the project ‘Trustworthy AI Tools for the Prediction of Obesity-Related Vascular Diseases (AI-POD)’ is to develop and validate AI tools for the assessment and prediction of risk of cardiovascular diseases and related complications in obese persons. AI-POD will advance the current state-of-the-art by introducing a novel, ground-breaking cardiovascular risk prediction and management approach for the obese population.
For the first time, all these factors will be considered in a novel, fair and trustworthy AI-based preventive risk score, integrating the influence of diet and lifestyle in real-time monitoring. AI-POD’s main outcomes are (1) a novel AI-based Clinical Decision Support System (CDSS) for the risk assessment (AI-POD risk score) and prediction of obesity-related CVDs and associated complications; (2) an innovative, easy-to-use mobile app for citizens (Citizen App) that interacts with the CDSS and considers lifestyle information.
During the first phase of the project, AI-POD made considerable progress. Notably, the AI-POD platform, which serves as the backbone for the data collection of the AI-POD clinical study, was created and the first machine learning (ML) models for heart CT scan analysis were developed. Also, an optimised photon counting CT protocol was established, the main goal of which is to gain the maximum possible diagnostic information. Moreover, the AI-POD Citizen App prototype was successfully launched. The app, linked to a fitness tracker, is now in use as part of the AI-POD prospective study to collect lifestyle information (e.g. daily activity level, dietary habits) and physical measurements such as BMI in addition to imaging data. These steps prepared the start of the prospective clinical study. The first patients were enrolled during this period, marking the beginning of the real-world application of the AI-POD results. Furthermore, the project made strides in designing the core functions of the CDSS, and ethical considerations were addressed by developing an overview of frameworks for guiding the responsible use of the AI-POD tools.
These achievements during the first project period highlight the significant progress of the AI-POD project towards its main objectives to develop and validate AI tools for the assessment and prediction of CVD risk.
These achievements during the first project period highlight the significant progress of the AI-POD project towards its main objectives to develop and validate AI tools for the assessment and prediction of CVD risk.
AI-POD works to demonstrate the potential of AI for integrating clinical, laboratory, and imaging data to enhance the assessment and management of obesity-related CVD. Our initial experiments support our hypothesis that the imaging-based AI-driven risk score and CDSS developed in AI-POD can significantly improve risk assessment and prediction of obesity-related CVD and its complications. The AI-POD risk score will be based on AI models connecting an unprecedented range of individual patient data, from cutting-edge highly resolved imaging, clinical and laboratory data to day-to-day dietary, activity and lifestyle information. It therefore has the potential to overcome current shortcomings of CVD diagnostics in obese persons and potentially transform the standard-of-care to the next level.
Furthermore, the AI-POD Citizen App has been set up and will empower individuals with obesity to take a more active role in monitoring and managing their health, while providing clinicians with more efficient workflows. These advancements not only have the potential to reduce morbidity and mortality among obese individuals but also highlight the capability of AI to ease public health burdens by optimising prevention and treatment strategies.
Furthermore, the AI-POD Citizen App has been set up and will empower individuals with obesity to take a more active role in monitoring and managing their health, while providing clinicians with more efficient workflows. These advancements not only have the potential to reduce morbidity and mortality among obese individuals but also highlight the capability of AI to ease public health burdens by optimising prevention and treatment strategies.