Center for Digital Health

2024 Pilot Grant Awardees

In October 2024, the Stanford Center for Digital Health announced the recipients of the annual CDH Pilot Grant Awards. These one-year grants of $50,000 were awarded to projects relating to digital health in the following research areas:

  • Ethical and responsible use of AI in health
  • Digital solutions for healthy aging, cognition, and longevity
  • Understanding the positive and/or negative impact of digital tools and services on health and wellbeing

Learn more about the selected projects below! 

Digital Health Innovations to Prevent Health Deterioration in Homeless Veterans

Contrary to popular perception, most homeless persons own a Smartphone or tablet and are amenable to using these devices to receive a range of care services. Despite this, we know little in the homeless population about (1) which individuals engage with digital health resources, and (2) the potential for device data in this population to facilitate real-time monitoring of health status. Drs. Blonigen and Zulman’s project will seek to answer these questions, with a broader goal of establishing a research program on digital phenotyping in homeless persons.

Daniel Blonigen, PhD, MA (PI) and Donna Zulman, MD, MS (Co-PI)

Building Trustworthy Digital Health Solutions: Using Crowdsourced Artificial Intelligence for Global Pediatric Healthcare

Drs. Wall and Darmstadt aim to integrate a crowd- and AI-powered autism spectrum disorder (ASD) detection mechanism into GuessWhat, a gamified mobile platform developed by theirlab. This will provide families around the world with a cheap, fast, and reliable tool to identify ASD (and, eventually, other pediatric mental health conditions) and receive the standard of care. To this end, the present proposal will 1) use GuessWhat to obtain a large library of videos of Bangladeshi children at risk for ASD, 2) recruit a local crowd workforce to label the resulting data and process it through the enhanced AI models, and 3) Train advanced AI for automatic feature extraction that speeds detection and expands reach to a global population of children with risk for DD while reducing bias.

Dennis Wall, PhD (PI) and Gary Darmstadt, MD, MS (Co-PI)

Informing ethical and responsible implementation of ambient AI scribe technology for diverse patient groups

This project aims to assess the impact of ambient AI scribe technology on patient care equity and privacy at Stanford Health Care, focusing on diverse patient groups and exploring patient perspectives. By employing a combination of quantitative analyses and qualitative methods, the study will identify potential disparities in the use of the AI tool and gather insights into patient autonomy and preferences. The ultimate goal is to inform the responsible implementation of AI-assisted documentation in healthcare settings, ensuring technologies like these enhance patient care across diverse patient groups.

Shreya Shah, MD (PI), Danton Char, MD (Co-PI)

Adolescent Health & Social Media: Measuring Outcomes of Digital Interventions

Excessive smartphone usage among adolescents is linked to negative outcomes such as poor sleep quality, increased anxiety, and decreased academic performance. By partnering with the one sec application, which has over 2.5 million users, the study aims to investigate digital interventions and correlate them with measurable health outcomes such as sleep quality, physical activity, and mental well-being. This pilot study will not only measure the impact of these interventions on health metrics but also identify the most effective types, providing essential data to support future research and funding applications aimed at improving adolescent health and well-being.

Paul Schmiedmayer, PhD (PI) and Emily Fox, PhD (Co-PI)

Automatic scoring of human speech recognition tests

This project will study the use of modern AI-based speech recognition to automatically score human speech recognition tests and identify patients with hearing loss who are at risk for cognitive impairment. The research will apply a machine-learning model to the additional data obtained by automatic scoring to identify meaningful information for characterizing patient performance, and detect those at risk for disorders like cognitive impairment. This project extends conventional speech in noise measures to conversational speech, allowing the researchers to characterize the effect of noise and cognitive impairments on a subject’s ability to enjoy the world around them.

Matthew Fitzgerald, PhD, MS (PI) and Malcolm Slaney, PhD, MSEE (Co-PI)

Real-time Melatonin Measurement for Studying Dim Light Melatonin Onset

Real-time melatonin measurement allows access to individual circadian rhythms, which is critical for personalized medicine and preventing neurodegeneration in the aging population. Current lab-based melatonin measurement methods often result in a time lag between sample collection and assessment, posing challenges for timely melatonin-based treatments. The objective of the project is to bring together expertise in sensors, materials, wireless communication, and physiological studies to develop a real-time melatonin sensor operable at home, explore the clinical significance of such a sensor and provide a digital solution for health monitoring in the aging population.

Ada S. Y. Poon, PhD, MS (PI), Makoto Kawai, MD (Co-PI)

Transforming the My Heart Counts smartphone app into a digital health platform for clinical trials

The My Heart Counts (MHC) smartphone app has broad reach with ~100K users and continues to publish high-impact digital health research. The team will integrate a pre-trained large language model to automatically generate coaching interventions aimed at increasing physical activity, which have proven to be as effective aso expert-curated coaching prompts in our preliminary data, in order to revolutionize the MHC app to serve as an autonomous digital health platform. With the integration of the trained large language model, the team will perform numerous fully digital clinical trials of physical activity in different disease contexts, with the aim of providing an evidence base for digital health interventions in improving human health and longevity.

Dan Seung Kim, MD, PhD, MPH (PI) 

Performance of a Medical Foundational Model for Central-line Insertion Site Assessments

In order to prevent central line-associated bloodstream infections (CLABSIs), current practice is for nurses to visually inspect and document the central line insertion site once per shift. This is a time-intensive and subjective process which could potentially be improved using artificial intelligence (AI), saving time for other direct patient care activities. This collaborative team previously trained a Medical Foundational Model (MedFM) with an image-text understanding using 4.5 million image-text pairs, with a general understanding of medical images and terminology surpassing human expert-level performance in various medical disciplines, including infectious diseases and dermatology. Building on these successes, the team aims to leverage MedFM’s advanced capabilities and explore its performance for surveillance and prevention of catheter-related bloodstream infections evaluating the catheter insertion site.

Jorge Salinas, MD (PI), Olivier Gevaert, PhD, MS (Co-PI)