Artificial intelligence-assisted analysis of efficacy, safety, and efficiency of pegylated factor VIII concentrates in hemophilia A based on product inserts

Manuel Rodríguez-López, Servicio de Hematología y Hemoterapia, Hospital Universitario Álvaro Cunqueiro, Instituto de Investigacion Sanitaria Galicia-Sur (IISGS), EOXI Vigo, Vigo, Pontevedra, España
Michael Calviño-Suárez, Servicio de Hematología y Hemoterapia, Complejo Hospitalario Universitario de Santiago de Compostela, Santiago de Compostela, La Coruña, España
Ramiro Núñez-Vázquez, Unidad de Trombosis y Hemostasia, Hospital Universitario Virgen del Rocío, Sevilla, España
Raquel Ocampo-Martínez, Servicio de Hematología y Hemoterapia, Hospital Universitario Álvaro Cunqueiro, Instituto de Investigacion Sanitaria Galicia-Sur (IISGS), EOXI Vigo, Vigo, Pontevedra, España

Introduction: There are different pegylated factor VIII concentrates, with different characteristics. Objective: To evaluate the use of basic AI, specifically ChatGPT-4o (OpenAI^®), to compare different pegylated CFVIII products due to the lack of direct comparative studies and the limitations of indirect comparisons. Material and method: Adynovi^®, Jivi^® and Esperoct^® are compared from their data sheets, with calculations based on an average patient weighing 70 kg. ChatGPT4o performs statistical analyses to compare efficacy (annualised bleeding rate [ABR]) and factor consumption. Results: ChatGPT4o suggests that Esperoct^® (50 IU/kg, 2 times/week) is the most cost-efficient regimen by annualised consumption and low ABR. Adynovi^®shows a slightly favourable safety profile, with no statistically significant differences in ABR or annual factor consumption (p > 0.05) between regimens, suggesting similar efficacy between the options. Conclusions: Although matching-adjusted indirect comparisons are validated tools for indirect comparisons in haemophilia, they face limitations. ChatGPT4o could help address them without replacing clinical validation. Artificial intelligence, including ChatGPT4o, could improve the accessibility and accuracy of matching-adjusted indirect comparisons, but still requires ongoing validation.

Keywords: Generative artificial intelligence. ChatGPT. Pegylated FVIIIC. Hemophilia A.