Background: Endometriosis is a chronic gynaecological disease of women of reproductive age linked to menstruation and the hormones that trigger it. The symptoms of endometriosis have a major impact on women's quality of life, either on a psychological level or on a societal one. Today, as many other female-specific topics, endometriosis is under-researched and under-diagnosed and many causes remain unknown. Although literature suggests that endometriosis patients are prone to use social network as a self-management tool and discuss various topics related to their disease, very few papers applied infodemiological (epidemiology based on internet health-related content) methods to endometriosis.

The main objective of this study is to understand whether infodemiology could be useful for public health decision makers in the management of patients with endometriosis and to prove it to be a reliable study methodology.

Methods: Firstly, social media post were collected between January 2020 and 2022 from France geolocalized sites using keyword for endometriosis. Automatic natural language processing methods were used to identify relevant patients with endometriosis posts expressing difficulties and unmet needs. Then, semi-structured interviews were conducted during July 2022 with 9 women with endometriosis and healthcare professionals. Discussions were recorded and transcribed verbatims were analysed through a thematic analysis allowing to generate general statistics on the most discussed topics.

Results: A total of 2,396 written by 1,742 patients with endometriosis were included. 2,317 (96.70%) were identified as containing at least one difficulty. The top 3 unmet needs and difficulties were difficulties present throughout the care pathway (71.1%; n = 1701), pathology-related (57.2%; n = 1,356) and treatments and medical procedures (20.6 %; n = 492). Semistructured interviews show that the top 3 main difficulties for patients with endometriosis were difficulties related to treatments (30% of the detected medical concepts), impact of the disease on the patient's environment (29% of the concepts) and transversal difficulties (16% of the concepts).

The bi-patient-centric approach with limited manual work allows to identify difficulties encountered by patients with endometriosis in France. The mixed-methodology of this study appears ideal to demonstrate their complementarity and to justify the use of AI algorithms for public health decision-making. Furthermore, endometriosis is a disease affecting a wide range of women, each with their own specific difficulties. They now expect societal and medical recognition of their illness and care pathways adapted to their individual symptoms.