Artificial intelligence and technology collaboratories: Empowering innovation in <scp>AI</scp> + <scp>AgeTech</scp>

Abstract

Launched in September 2021, the Artificial Intelligence and Technology Collaboratories (AITC) for Aging Research program is the newest of the seven centers programs funded by the National Institute on Aging (NIA), a part of the National Institutes of Health (NIH), and is dedicated to helping Americans live longer, healthier lives through the application of AI and emerging technologies (Figure 1). Three collaboratories, centered at Johns Hopkins University (JH AITC), the University of Massachusetts Amherst (MassAITC), and the University of Pennsylvania (PennAITech), and a Coordinating Center managed by Rose Li & Associates Inc. (RLA) comprise the “a2 Collective.” By committing more than $65 million over 5 years toward this program, NIA amplified the promise of technology, and AI in particular, to accelerate the development of solutions to help Americans, especially those with dementia, live where they most want: in their homes.1 NIA relies on proven tactics for the greatest impact: bringing together multiple disciplines to tackle societal challenges, cultivating timely data sharing where possible, identifying and enticing top talent in varied fields to focus on aging, and orchestrating a harmonized approach from the start, with national scope, to give the program staying power in a fast-evolving AgeTech ecosystem. NIA earmarked $40 million over 5 years for the a2 Collective to hold annual pilot award competitions. Our first three calls for applications drew nearly 700 applications from 45 states plus Washington, DC, Puerto Rico, and the U.S. Virgin Islands. A significant majority (71%) of the initial 60 awardees (from two competitions) include academic collaborators, reflecting the premium placed on research rigor (Figure 2). More than 40% of pilots are led by women, and about three-quarters relate to dementia. Specific examples include a machine learning-enabled, speech-based dementia screening tool for families and caregivers and a simple imaging and telemedicine system for remote eye (cataract) screening in disadvantaged populations by non-ophthalmologists. (See additional funded pilot descriptions at https://www.a2collective.ai/awardees.) Most pilot projects are developing or beta testing prototypes (60%) and evaluating prototypes in real-world conditions (30%); fewer are at the technology concept or discovery stage (7%) or pursuing commercial deployment or scaling up (3%). About 80% of funded pilots involve machine learning (ML) and significant proportions are developing user-facing software and platforms, wearables, smart household devices or utilities, and environmental sensors. Pilot awards to date were selected from an application pool constituted before the advent of widespread public engagement with powerful large language models (LLMs) at the end of 2022. Given the many LLMs now available, future calls are expected to see a dramatic increase in the number of projects building on generative AI and LLMs. The landscape is becoming saturated with technologies and platforms with overlapping functions. Comparative effectiveness research is increasingly needed to facilitate inevitable consolidation in the field. In the meantime, we need to continue to identify and support the most innovative technology and partnerships, invest in projects that take calculated risks and employ ethical AI design, pursue practical product-development goals, and demonstrate an understanding of the user experience and integration into clinical practice. Almost all awarded pilots are collecting human subjects data, but generally in small numbers. Giving more weight to projects that access large datasets (e.g., through academic research institutions, industry, or payers) and prioritizing the completeness of data will accelerate the development of more sophisticated, inclusive AI/ML methods. The pace of AI evolution may simply be too rapid for our pilot projects to reflect the current research frontier in real time. Reducing the time to project launch would help our pilots more nimbly respond to fast-changing developments. NIA deserves credit for entering the AgeTech space in a meaningful way, building on its earlier investments in research on technology use among older adults and using AI and technology to improve older adults' health and well-being.2 Because of NIA support, the a2 Collective is contributing by bringing greater academic rigor to research underpinning innovations and sharing advances that are helping to shape the fast-moving field. All authors contributed to reviewing and concurring with the manuscript content prior to submission. RML led the writing of the manuscript. We thank Partha Bhattacharya, National Institute on Aging (NIA), for the vision and leadership he has contributed to the AITC program, and Kayla Harr for editorial assistance with the manuscript. This work was supported by grants U24AG073094 (RML, KH), P30AG073104 (PMA, AB, RC, JDW), P30AG073107 (NKC and DG), and P30AG073105 (GD, JK, and JHM) from the NIA, part of the National Institutes of Health. No conflict of interest exists for any of the authors.

Ähnliche Arbeiten