Clinical decision-making in benzodiazepine deprescribing by healthcare providers vs. AI-assisted approach
Само за регистроване кориснике
2024
Аутори
Bužančić, IvaBelec, Dora
Držaić, Margita
Kummer, Ingrid
Brkić, Jovana
Fialová, Daniela
Ortner Hadžiabdić, Maja
Чланак у часопису (Објављена верзија)
Метаподаци
Приказ свих података о документуАпстракт
Aims: The aim of this study was to compare the clinical decision-making for benzodiazepine deprescribing between a healthcare provider (HCP) and an artificial intelligence (AI) chatbot GPT4 (ChatGPT-4). Methods: We analysed real-world data from a Croatian cohort of community-dwelling benzodiazepine patients (n = 154) within the EuroAgeism H2020 ESR 7 project. HCPs evaluated the data using pre-established deprescribing criteria to assess benzodiazepine discontinuation potential. The research team devised and tested AI prompts to ensure consistency with HCP judgements. An independent researcher employed ChatGPT-4 with predetermined prompts to simulate clinical decisions for each patient case. Data derived from human-HCP and ChatGPT-4 decisions were compared for agreement rates and Cohen's kappa.
Results : Both HPC and ChatGPT identified patients for benzodiazepine deprescribing (96.1% and 89.6%, respectively), showing an agreement rate of 95% (κ=.200,P=.012). Agreement on four deprescr...ibing criteria ranged from 74.7% to 91.3%(lack of indicationκ=.352,P< .001; prolonged useκ=.088,P=.280; safety concernsκ=.123,P=.006; incorrect dosageκ=.264,P=.001). Important limitationsof GPT-4 responses were identified, including 22.1% ambiguous outputs, genericanswers and inaccuracies, posing inappropriate decision-making risks.Conclusions : While AI-HCP agreement is substantial, sole AI reliance poses a risk forunsuitable clinical decision-making. This study's findings reveal both strengths andareas for enhancement of ChatGPT-4 in the deprescribing recommendations within areal-world sample. Our study underscores the need for additional research on chatbot functionality in patient therapy decision-making, further fostering the advance-ment of AI for optimal performance.
Кључне речи:
artificial intelligence (AI) / benzodiazepines / chatbot / ChatGPT-4 / deprescribingИзвор:
British Journal of Clinical Pharmacology, 2024, 90, 3, 662-674Издавач:
- John Wiley and Sons Inc
Финансирање / пројекти:
- The EuroAgeism H2020 project
DOI: 10.1111/bcp.15963
ISSN: 0306-5251
PubMed: 37949663
Scopus: 2-s2.0-85178459514
Институција/група
PharmacyTY - JOUR AU - Bužančić, Iva AU - Belec, Dora AU - Držaić, Margita AU - Kummer, Ingrid AU - Brkić, Jovana AU - Fialová, Daniela AU - Ortner Hadžiabdić, Maja PY - 2024 UR - https://farfar.pharmacy.bg.ac.rs/handle/123456789/5344 AB - Aims: The aim of this study was to compare the clinical decision-making for benzodiazepine deprescribing between a healthcare provider (HCP) and an artificial intelligence (AI) chatbot GPT4 (ChatGPT-4). Methods: We analysed real-world data from a Croatian cohort of community-dwelling benzodiazepine patients (n = 154) within the EuroAgeism H2020 ESR 7 project. HCPs evaluated the data using pre-established deprescribing criteria to assess benzodiazepine discontinuation potential. The research team devised and tested AI prompts to ensure consistency with HCP judgements. An independent researcher employed ChatGPT-4 with predetermined prompts to simulate clinical decisions for each patient case. Data derived from human-HCP and ChatGPT-4 decisions were compared for agreement rates and Cohen's kappa. Results : Both HPC and ChatGPT identified patients for benzodiazepine deprescribing (96.1% and 89.6%, respectively), showing an agreement rate of 95% (κ=.200,P=.012). Agreement on four deprescribing criteria ranged from 74.7% to 91.3%(lack of indicationκ=.352,P< .001; prolonged useκ=.088,P=.280; safety concernsκ=.123,P=.006; incorrect dosageκ=.264,P=.001). Important limitationsof GPT-4 responses were identified, including 22.1% ambiguous outputs, genericanswers and inaccuracies, posing inappropriate decision-making risks.Conclusions : While AI-HCP agreement is substantial, sole AI reliance poses a risk forunsuitable clinical decision-making. This study's findings reveal both strengths andareas for enhancement of ChatGPT-4 in the deprescribing recommendations within areal-world sample. Our study underscores the need for additional research on chatbot functionality in patient therapy decision-making, further fostering the advance-ment of AI for optimal performance. PB - John Wiley and Sons Inc T2 - British Journal of Clinical Pharmacology T1 - Clinical decision-making in benzodiazepine deprescribing by healthcare providers vs. AI-assisted approach VL - 90 IS - 3 SP - 662 EP - 674 DO - 10.1111/bcp.15963 ER -
@article{ author = "Bužančić, Iva and Belec, Dora and Držaić, Margita and Kummer, Ingrid and Brkić, Jovana and Fialová, Daniela and Ortner Hadžiabdić, Maja", year = "2024", abstract = "Aims: The aim of this study was to compare the clinical decision-making for benzodiazepine deprescribing between a healthcare provider (HCP) and an artificial intelligence (AI) chatbot GPT4 (ChatGPT-4). Methods: We analysed real-world data from a Croatian cohort of community-dwelling benzodiazepine patients (n = 154) within the EuroAgeism H2020 ESR 7 project. HCPs evaluated the data using pre-established deprescribing criteria to assess benzodiazepine discontinuation potential. The research team devised and tested AI prompts to ensure consistency with HCP judgements. An independent researcher employed ChatGPT-4 with predetermined prompts to simulate clinical decisions for each patient case. Data derived from human-HCP and ChatGPT-4 decisions were compared for agreement rates and Cohen's kappa. Results : Both HPC and ChatGPT identified patients for benzodiazepine deprescribing (96.1% and 89.6%, respectively), showing an agreement rate of 95% (κ=.200,P=.012). Agreement on four deprescribing criteria ranged from 74.7% to 91.3%(lack of indicationκ=.352,P< .001; prolonged useκ=.088,P=.280; safety concernsκ=.123,P=.006; incorrect dosageκ=.264,P=.001). Important limitationsof GPT-4 responses were identified, including 22.1% ambiguous outputs, genericanswers and inaccuracies, posing inappropriate decision-making risks.Conclusions : While AI-HCP agreement is substantial, sole AI reliance poses a risk forunsuitable clinical decision-making. This study's findings reveal both strengths andareas for enhancement of ChatGPT-4 in the deprescribing recommendations within areal-world sample. Our study underscores the need for additional research on chatbot functionality in patient therapy decision-making, further fostering the advance-ment of AI for optimal performance.", publisher = "John Wiley and Sons Inc", journal = "British Journal of Clinical Pharmacology", title = "Clinical decision-making in benzodiazepine deprescribing by healthcare providers vs. AI-assisted approach", volume = "90", number = "3", pages = "662-674", doi = "10.1111/bcp.15963" }
Bužančić, I., Belec, D., Držaić, M., Kummer, I., Brkić, J., Fialová, D.,& Ortner Hadžiabdić, M.. (2024). Clinical decision-making in benzodiazepine deprescribing by healthcare providers vs. AI-assisted approach. in British Journal of Clinical Pharmacology John Wiley and Sons Inc., 90(3), 662-674. https://doi.org/10.1111/bcp.15963
Bužančić I, Belec D, Držaić M, Kummer I, Brkić J, Fialová D, Ortner Hadžiabdić M. Clinical decision-making in benzodiazepine deprescribing by healthcare providers vs. AI-assisted approach. in British Journal of Clinical Pharmacology. 2024;90(3):662-674. doi:10.1111/bcp.15963 .
Bužančić, Iva, Belec, Dora, Držaić, Margita, Kummer, Ingrid, Brkić, Jovana, Fialová, Daniela, Ortner Hadžiabdić, Maja, "Clinical decision-making in benzodiazepine deprescribing by healthcare providers vs. AI-assisted approach" in British Journal of Clinical Pharmacology, 90, no. 3 (2024):662-674, https://doi.org/10.1111/bcp.15963 . .