AI-Driven Social Listening Research on Menopausal Symptom Burden and Impact on Quality of Life in Women With Natural or Induced Menopause

OBJECTIVES: This study aims to explore the feasibility to characterize targeted cohorts of women with menopausal symptoms participating in online communities and to assess how specific aspects, e.g., symptom profile and severity, impact on Quality of Life (QoL), can be captured via retrospective Artificial Intelligence (AI)-driven social listening.

METHODS: Posts from thirteen online health communities (US, UK) were retrieved between April 2020-April 2023. Women’s experiences were algorithmically coded using dedicated NLP analyzers to detect symptoms, quality of life (QoL) facets, treatments, as well as identify cohorts of natural menopause (NM) or induced menopause (IM). Women who self-reported menopause cause as, e.g., surgery or adjuvant endocrine therapy were grouped under IM, otherwise under NM cohort. Impact of symptoms was measured by an NLP analyzer capturing impairment relations between symptoms and an affected QoL facet as self-reported by women. Data were collected and processed in an anonymized and secure way to ensure privacy and ethical aspects.

RESULTS: 154,704 posts authored by 13,953 menopausal women were collected. In a subset of 3,177 women with NM and 255 with IM, symptom burden and impact on QoL was evaluated in more detail. Age was algorithmically determined for 29% of women (N=4,008, median: 49 years). Most frequently reported symptoms in the cohorts were vasomotor symptoms (53% in NM and 65% in IM cohort), pain (NM: 45%, IM: 59%), mood changes (NM: 53%, IM: 50 %), fatigue (NM: 44%, IM: 38%), and sleep disturbances (NM: 26%, IM: 26%). Among QoL facets included in the study, ‘Sleep and rest’ was mentioned most frequently as being affected.

CONCLUSIONS: Social listening was identified as a feasible approach to gather first-hand insights from women participating in online communities about their menopausal symptoms. Our results indicate that menopausal symptoms have a markedly high burden on individual QoL aspects, strengthening similar findings from the literature.