AI chatbot answers for drug dosing adjustments according to renal function in geriatric patients: A cross-sectional study using the new scoring system AQUOS (AI-Quality-Output-Score)
ABSTRACT
Background:
Preventable adverse drug reactions (ADRs) in geriatric patients are caused by overdosing, especially with impaired renal function. AI (artificial-intelligence) chatbots are discussed as tools to generate drug information, which can adjust drug dosing and prevent subsequent ADRs based on individualized patient data. However, the question arises as to what extent can such AI chatbots withstand scientific evaluation in this task.
Objective:
We newly developed and validated the AI-Quality-Output-Score (AQUOS, ranging from 0% to 100%) to assess the quality of AI chatbot answers. We investigated whether AQUOS depends on (i) renal function, (ii) medication complexity, (iii) prompting language English-German, and (iv) whether the answers are reproducible (assessment at two independent times). Additionally, we assessed the potential harm.
Methods:
In a standardized prompt, we asked four AI chatbots whether the medication of 100 geriatric patients with polymedication at discharge should be adjusted according to their renal function. We prompted drug-related queries in two languages and at two times to assess AI chatbot answers, and we scored the generated outputs based on AQUOS. Additionally, we assessed possible harm of the AI chatbot answers using the WHO-definition “The-conceptual-framework-for-the-international-classification-for-patient-safety”.
Results:
We analyzed 1,600 AI chatbot answers, with AQUOS values ranging from -19.0 to 95.2%, depending on chatbot. We found that AQUOS declined with decreasing renal function (i) (ChatGPT -0.215, p = .03) and increasing medication complexity (ii) (scite -0.239, p = .02). Possible harm also correlated with more complicated patient statuses (lower kidney function and higher medication complexity) across all chatbots. Overall scores were up to 4.8% higher in English than in German prompting (iii). The AI chatbot answers were highly reproducible (iv).
Conclusions:
In renal drug dosing, the quality of AI chatbot answers declined with decreased renal function and increased medication complexity. Even the highest AQUOS achieved is insufficient for deploying AI chatbots in the high-risk healthcare sector.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.