From Lived Experience to Evidence | AI-Driven Social Media Listening, Predictors, and Everyday Outcomes in Blood Cancers

With September dedicated to recognising the impact of research, studies, science, patients, carers, patient organisations, innovators, policymakers, and more in the blood cancers, we are taking a moment to share what our community has achieved and where we’re heading next. In the past 12 months, we have really focused on bringing more innovative research, technologies, and ways of engaging with patients, carers, patient organisations and clinical colleagues, with a strong focus on having a real, positive impact and improving outcomes for patients.

What we’ve seen across all indications over the past 10 years, and more acutely in the last 12 months, is a real need for stronger insight into patient-reported outcomes, day-to-day feedback, and continuously captured datasets. To that end, we have really doubled down on embracing digital health tech approaches, combining patient-reported data with passively captured biometric signals from wearables and AI-driven analytics to generate some deep insights that support patients and carers, advance research, strengthen industry’s understanding of unmet need through real-world evidence (RWE), and inform clinical care pathways. Together, these aspects centre on helping people to stay well at home, with better visibility of their own health.

Our Work Across Blood Cancers and Haematology

Across Chronic Lymphocytic Leukaemia (CLL), our work published in Blood as part of the 2024 ASH Annual Meeting demonstrated the use of large-scale AI social media listening (SML) to understand real-world experiences directly from patients and carers. Through web scraping of sources such as Instagram, TikTok, Twitter, YouTube, Google Reviews, Reddit, and online articles, followed by AI-powered sentiment and theme detection, we surfaced some of the key things that matter most to the CLL community.

Overall sentiment found that, by and large, these posts expressed a hopefulness about treatment progress, yet a persistent mental load during “watch and wait”. Carer burden was the most negatively scored theme, while community support was the most positive. These theme-level details furthermore highlighted “treatment and management” as the most frequent topic, with the majority positive. Meanwhile, “disease knowledge and awareness” was dominated by neutral queries, highlighting a clear need for accessible information, and significant negative sentiment in “carer experiences”. These AI-derived insights are guiding targeted education, active monitoring support between visits, and patient-centred engagement across the CLL community.

In Multiple Myeloma (MM), we integrated longitudinal primary care data with prospective, patient-led digital capture to characterise disease burden and everyday outcomes. These retrospective PC records mapped events, medications and comorbidities across patients’ medical histories, while a prospective cohort contributed almost 40,000 ePRO datapoints across key quality of life-linked metrics and continuously captured wearable vitals.

With early differences by sex, age and treatment line evident, these findings, due to be presented in detail at IMS 2025 later this year, demonstrated the feasibility of exploring routine records alongside at-home outcomes and physiology without adding burden to clinical teams, while keeping patients in control. Headline signals established cohort baseline means in key validated measures, such as the EQ-5D-5L index and Health State scores, while highlighting key symptoms with the greatest severity, such as weakness and drowsiness. With a number of meaningful variations in vitals such as night-time SpO₂ and breathlessness by treatment line, these insights support more tailored community-based support and prioritisation of what to track between visits.

Outside of blood cancers, our work in social media listening has also extended to Sickle Cell Disease (SCD), in partnership with industry, which continues to underscore day-to-day realities in the UK. Key barriers were in urgent care, the search for empathy, and the work patients do to understand triggers, pain and fatigue, with a critical call for better access to effective treatments. These voices directly shape where we focus our measurements for tracking and wraparound support for the community.

From SML to outcomes, the upcoming ASCAT 2025 congress will see us present an oral iteration of our EHA 2025 work, demonstrating how daily ePROs and wearable biometrics illuminate a pre-vaso-occlusive (VOC) window and stratify risk before escalation. In our consented UK patient cohort, we evaluated variation across key metrics across the seven-day pre-VOC period in comparison to baseline and VOC days themselves. Of note, EQ-5D-5L and Health State scores dropped as early as one day before VOC, and hour-by-hour wearable trends showed lower nocturnal respiratory rate and SpO₂ ahead of and during VOCs versus baseline.

Moreover, 18% of the self-reported VOCs involved a hospitalisation, with pre-hospitalisation patterns featuring longer moderate/soft activity duration, higher step counts, and elevated pulse rates versus non-hospitalised VOCs, alongside sharper drops in activity and sleep during hospitalised VOCs. These differences point to practical AI-driven risk stratification that will enable earlier, non-clinical interventions to help people stay safer at home – supporting the feasibility of future remote VOC prediction strategies through key physiological signals in the pre-VOC period.

As we reflect on these findings, our focus is on turning evidence into everyday support. Each day, we continue to work alongside patients, clinicians, and partners to translate real-world voice, day-to-day signals, and AI insights into practical support, clearer pathways and stronger evidence for decisions. The aim is simple: better visibility for people over their own health, and the support to stay well at home while services concentrate where they can have the greatest impact.

For patients, join our mission. For clinicians, thank you for your continued support. For industry, thank you for your engagement.