We are writing in response to the editorial “AI and Technology in Geriatrics: A New Chapter in JAGS,” by Abadir and Schlesinger published in the Journal of the American Geriatrics Society 1. Artificial Intelligence (AI) may represent the most important intervention to address the seemingly unsolvable barriers that limit the access of older adults to timely, comprehensive, and individualized health care. The creation of the new section of the Journal of the American Geriatrics Society (JAGS): JAGS-AI & TECH is vitally needed as a scholarly platform for the multidisciplinary review and dissemination of emerging AI interventions 1, 2. We strongly agree with Abadir and Schlesinger that AI will not replace geriatricians 1. Rather, AI is the gamechanger that can potentially expand the accessibility and effectiveness of geriatricians for an ever-growing number of older adult patients. Collaborations between technology companies, healthcare systems, and AI researchers have been the primary drivers for AI product development. However, we agree that geriatricians have the unique skills to provide the “clinical insight” and “ethical stewardship” for the development, implementation, and assessment of future AI healthcare platforms affecting older adults 1. For example, an AI model that uses cognitive screening scores to “label” older adults with cognitive impairment may need geriatrician guidance to ensure that factors related to age, level of education, language proficiency, mood disorders, and sensory impairment are also taken into consideration. Other instances where geriatrician expertise would be indispensable include calibrating fall-risk models, developing deprescribing algorithms, and monitoring cognitive and behavioral changes in the home. Table 1 expands Abadir and Schlesinger's future roadmap 1 by highlighting the unique role for geriatricians in this emerging AI landscape. Artificial intelligence models can continuously refine the analysis of existing predictive comorbidity indices to develop ever-improving patient specific predictions 3. Opportunities for health service research will emerge as deep learning models help uncover new associations for geriatric syndromes and monitor the impact of clinical and population health interventions. Geriatricians must play a role in assessing the quality of data as an age bias can occur when AI models do not represent the age heterogeneity of the overall geriatric population. This modeling must also take into account racial and cultural biases to ensure that historically underrepresented populations (i.e., Black, Latinx, and LGBTQ+) are adequately represented in global datasets 4. Failure to do so could lead to models that make incorrect or even dangerous assumptions about individual geriatric care patients. The capital and maintenance expenses for a comprehensive geriatric health AI platform would likely come from a large payor source(s). The measurement of the return on investment for AI systems should have geriatrician input to assess the impact on patient quality indicators, such as avoidable hospitalizations and delayed nursing home placement. While current AI efforts are attempting to decrease physician documentation burden, more advanced AI systems could operate with increasing levels of autonomy (Agentic AI) to perform real-time medical decision-making tasks that are linked to specific physician orders. Similarly, AI models could also be trained to help clinicians match data regarding patient-specific conditions, functional impairments, and current health care entitlements with the just-in-time local availability of health care resources. AI models can also be developed to guide geriatricians to make appropriate selections among the wide milieu of wearable and home health monitoring devices. We look forward to the impact that JAGS-AI & TECH will have on improving the quality of geriatric-focused AI initiatives. All authors contributed to the manuscript. The authors have nothing to report. This material is the result of work that was supported with resources and the use of facilities at the Orlando VA Healthcare System. The contents do not represent the views of the Department of Veterans Affairs or the United States Government. The authors declare no conflicts of interest. This publication is linked to a related reply by Peter M. Abadir. To view this article, visit https://doi.org/10.1111/jgs.70508.
Golden et al. (Sat,) studied this question.