Artificial Intelligence (AI)
Artificial intelligence (AI) is a crosscutting technology that connects and amplifies the Duke Center for Computational and Digital Health Innovation’s three core capabilities—wearable sensing, high performance computing (HPC), and extended reality.
Through AI for engineering, we harness machine learning, physics-informed modeling, and real-time analytics to bridge these technologies in powerful and novel ways. This integration is central to how we build intelligent systems that don’t just analyze data—they simulate, predict, and interact. By embedding AI into computational models and immersive tools, we are uniquely positioned to find emerging disease risks, track patient health over time, and treat illness with personalized precision—advancing our mission to Find, Track, and Treat.
From federated learning algorithms scaling across HPC platforms to AI-enhanced digital twins for personalized care, AI is the engine driving discovery and transformation in computational health.
What Is AI?
Artificial intelligence refers to a diverse set of computational techniques that enable machines to learn from data, recognize patterns, make decisions, and support human reasoning. This includes machine learning, deep learning, natural language processing, and neural networks—methods that continue to evolve rapidly in both capability and accessibility.
At the Center, we go beyond applying AI as a standalone tool. We embed it within simulation frameworks, interface systems, and large-scale computing environments to interpret complex multimodal health data, automate time-intensive analysis, and drive real-time, patient-specific decision support. This integrated approach to AI allows us to create adaptive and intelligent tools that are not only accurate but also actionable in clinical and research settings.
What Are the Advantages of AI?
AI can help us revolutionize healthcare through:
- Data analysis: Identifying early signs of disease from diverse, high-dimensional datasets
- Predictive modeling: Using algorithms and statistical techniques to analyze current and historical data to make predictions about future outcomes.
- Real-time simulations: Enabling faster, more adaptive, and more accurate modeling of complex biological systems.
- Acceleration of research: Enhancing the speed and scale of research by automating analysis of massive datasets.
How We’re Using AI
AI is deeply integrated into many of the projects at the Center for Computational and Digital Health. Paired with our driving technologies, AI will help us detect disease earlier, improve diagnosis, and personalize health care.
A few core examples are the following:
- The Calabrese Lab uses AI to create virtual models — digital twins or digital phantoms — to make a virtual model of the patients’ arterial anatomy for further analysis, especially as it relates to intracranial aneurysms. This allows them to simulate multiple CT scans of a patient’s brain and their aneurysm without having to scan the patient again.
- The Randles Lab developed HarVI (HARVEY Virtual Intervention), a groundbreaking AI-powered tool that integrates real-time hemodynamic analysis with extended reality for immersive, intuitive user interaction. HarVI enables real-time hemodynamic feedback using a predefined set of 1D CFD simulations for machine learning-based training, allowing instantaneous FFR predictions with an average processing time of just 74 minutes, compared to the 24-48 hours required by existing tools.om underserved communities. The lab has demonstrated that data from wearables can be integrated into electronic health records, supporting innovative care models that are both clinically effective and financially sustainable.
Research Featuring AI
-
Development of a Pipeline for Transitioning Wearable and Non-Wearable Data to a Centralized Storage, Analysis and Downstream Applied Research Platform
The Luck Lab developed an integrated data framework that links youth athletes’ reported activity and exposure with wearable head-impact kinematics to improve the ability…
-
-
-
Publications Featuring AI
-
Impact of daylight saving time on physical activity patterns
Daylight saving time (DST) remains contentious: some policymakers highlight behavioural benefits, while others emphasize health risks. Here we estimated the behavioural and physiological impacts…
-
Catch Smart Glasses If You Scan: Robot-Controlled vs. UI-Guided
As smart glasses become increasingly lightweight and suitable for all-day wearables, they show strong potential to serve as the next-generation mobile platforms that provide…
-
The need to develop health data transaction disclosure requirements to balance transparency, privacy, and progressive use
This Viewpoint critically evaluates the impact of commodifying health data in the era of electronic health records and the ethical challenges this practice raises.…
-
Translational Methods for Wearable Heart Rate Variability Monitoring in Older Adults: Preoperative Risk Stratification and Postoperative Monitoring
Low preoperative heart rate variability (HRV) is associated with adverse cardiovascular outcomes and delirium after geriatric surgery, but in-person testing is burdensome and impractical.…
-
WatchSleepNet: A Novel Model and Pretraining Approach for Advancing Sleep Staging with Smartwatches
Sleep monitoring is essential for assessing overall health and managing sleep disorders, yet clinical adoption of consumer wearables remains limited due to inconsistent performance…
-
Building an open-source community to enhance autonomic nervous system signal analysis: DBDP-autonomic
Smartphones and wearable sensors offer an unprecedented ability to collect peripheral psychophysiological signals across diverse timescales, settings, populations, and modalities. However, open-source software development…
-
A method for intelligent allocation of diagnostic testing by leveraging data from commercial wearable devices: a case study on COVID-19
Mass surveillance testing can help control outbreaks of infectious diseases such as COVID-19. However, diagnostic test shortages are prevalent globally and continue to occur…
-
Non-invasive wearables for remote monitoring of HbA1c and glucose variability: proof of concept
Diabetes prevalence continues to grow and there remains a significant diagnostic gap in one-third of the US population that has pre-diabetes. Innovative, practical strategies…
-
Engineering digital biomarkers of interstitial glucose from noninvasive smartwatches
Prediabetes affects one in three people and has a 10% annual conversion rate to type 2 diabetes without lifestyle or medical interventions. Management of…
-
Assessment of ownership of smart devices and the acceptability of digital health data sharing
Smart portable devices- smartphones and smartwatches- are rapidly being adopted by the general population, which has brought forward an opportunity to use the large…