Evidano home
Commentary on News

AI Qualitative Social Media Analysis for Cancer Research

Evidano9 min read

Healthcare researchers need to understand not only what cancer communities discuss, but also how patients, advocates, oncologists, media organizations, and industry interact over time. Qualitative social media analysis can uncover the lived experience, knowledge exchange, and policy concerns that frequency counts alone cannot explain. The November 20, 2026 study "Tweeting with precision: A multiscale longitudinal digital ethnography of online cancer communities" shows how a mixed-method approach can connect large-scale patterns with detailed conversations about biomarkers, targeted therapies, clinical trials, and access to care. This article translates that research into practical guidance for teams conducting AI qualitative research. Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze social media data, interviews, open-ended surveys, and research documents. The platform can help researchers move from a large text corpus to themes, cross-segment comparisons, and evidence-backed questions while preserving human review and research ethics.

Key Takeaways

According to Tweeting with precision: A multiscale longitudinal digital ethnography of online cancer communities, qualitative social media analysis shows that cancer communities use online platforms for scientific information, peer support, advocacy, and challenges to healthcare policy.

The study demonstrates that AI-assisted qualitative research is most useful when computational scale is combined with close reading of representative conversations, stakeholder classification, and interaction analysis.

  • The PLOS One study analyzed 8, 614, 284 Twitter publications from selected cancer-community accounts, including 4, 389, 619 tweets and 3, 031, 281 retweets, using data collected in March and April 2023 about activity spanning 2010 to 2022.
  • The researchers selected 834 highly active users from 28, 524 users, with the selected users representing less than 3% of users but producing more than 80% of tweets containing the relevant cancer hashtags.
  • The study identified 116 initial topic clusters, consolidated them into 49 topic groups and 13 overarching categories, with cancer topics accounting for 50% of the corpus after outliers were excluded.
  • The study found that 2.4% of tweets referred to precision oncology, while oncologists authored 40% of biomarker-related tweets and advocates and media each authored 17%.
  • The researchers analyzed 7, 670 tweets across approximately 3, 000 conversations involving Janet Freeman-Daily to examine how ROS1 advocacy developed through online and offline relationships.

How the Study Used Qualitative Social Media Analysis

The PLOS One study used a mixed-method design that combined semantic analysis, user classification, retweet-network analysis, and close qualitative reading of conversations.

The researchers began with approximately 30 cancer-related hashtags and then expanded the corpus by retrieving the complete tweet histories of 834 highly active users. The approach provided broad coverage while also allowing the researchers to inspect how specific actors discussed biomarkers such as ROS1, ALK, EGFR, KRAS, and HER2.

The researchers manually classified all 834 accounts into eight categories: patients, advocates, oncologists, health professionals, health centers, media, nonprofit organizations, and industries. Four researchers reviewed account handles and biographies, with ambiguous cases discussed until consensus was reached.

For semantic analysis, the study used BERTopic with transformer-based embeddings, UMAP dimensionality reduction, and HDBSCAN clustering. The researchers then reviewed the 50 most representative tweets for each topic to validate topic coherence and assign meaningful labels.

The study authors summarized the value of combining scale and interpretation directly: "We combine quantitative and qualitative approaches." The researchers explained that quantitative analysis showed interaction patterns and themes, while qualitative analysis revealed what those exchanges meant for patients, care organization, and clinical-trial regulation.

Findings Snapshot: Cancer Community Research in 2026

Date or periodMetricValueResearch implication
2010 to 2022Tweeting activity in the full dataset4, 389, 619 tweetsLarge corpora require automated thematic and frequency analysis before human interpretation.
2010 to 2022Initial topic clusters116Machine-generated clusters need qualitative review and consolidation into interpretable themes.
2010 to 2022Overarching topic categories13Hierarchical coding can connect detailed subtopics to broader research questions.
2010 to 2022Precision oncology share of tweets2.4%A small thematic segment can still reveal important discussions about targeted therapies and access.
2015ROS1ders group formationOnline and in-person mobilizationQualitative chronology can identify how digital interactions precede organized advocacy.
July 2013 to December 2021LCSM tweetchatsMore than 150 chatsRepeated conversations can be analyzed as an evolving community rather than isolated posts.

Implications for Healthcare and Policy Researchers

Why should researchers combine thematic analysis with network analysis?

Researchers should combine thematic analysis with network analysis because the PLOS One study found that discussion content and information circulation answered different parts of the research question.

Topic analysis showed what stakeholders discussed, including cancer awareness, scientific news, clinical trials, healthcare systems, and social interaction. Retweet analysis showed how advocates, oncologists, patients, and organizations connected, including the study's descriptive finding that advocates and oncologists tended to share content within their own groups.

What did the study reveal about patient advocates?

The study found that patient advocates acted as bridges between patients and medical professionals by sharing information from a broader range of sources than patients and by engaging with oncologists, researchers, and healthcare organizations.

The study reported that advocates tweeted twice as much as oncologists in proportion to their group size, although the authors cautioned that this was a descriptive observation rather than a causal finding. One advocate wrote, "Twitter had been ‘the one place where all stakeholders [...] could meet and exchange ideas on something that had life-and-death implications for so many.’"

How can qualitative analysis identify emerging research priorities?

Qualitative analysis can identify emerging research priorities by tracing how repeated conversations connect scientific findings with patient experience, treatment access, clinical-trial design, and regulation.

The PLOS One study's ROS1 case showed that online exchanges preceded the formation of The ROS1ders in 2015 and accompanied discussion of crizotinib, which received U.S. Food and Drug Administration approval for ROS1-positive metastatic non-small-cell lung cancer on March 11, 2016. The researchers concluded that online interactions could help make rare biomarkers and patient needs visible to researchers and health authorities.

What limitations should healthcare researchers document?

Healthcare researchers should document sampling, platform, language, privacy, and representativeness limitations because the PLOS One study focused on highly active accounts and a predominantly Anglo-American context.

The study noted that 834 accounts represented less than 3% of identified users, which may overrepresent committed or professional voices. The authors also reported that only tweet IDs were shared because of ethical, legal, and platform-policy concerns, and they warned that deletion, account suspension, and platform changes limit exact reproducibility.

How Evidano Helps With AI Qualitative Research

Problem: Large social media corpora are difficult to review manually

Solution: Evidano can organize large collections of documents and text responses for thematic, content, frequency, and cross-segment analysis, helping researchers identify recurring topics before selecting material for close reading.

Researchers can use hierarchical codes and subcodes to separate broad themes, such as cancer advocacy, from narrower topics, such as ROS1 treatment access or clinical-trial participation. Learn more about Evidano's research capabilities on the Evidano features page.

Problem: Stakeholder differences are easy to miss in aggregate results

Solution: Cross-segment analysis can compare themes across groups such as patients, advocates, clinicians, nonprofit organizations, and media accounts.

Segment comparisons can help research teams ask whether a topic appears across the whole dataset or is concentrated among one stakeholder group. Human researchers should still validate labels, review representative excerpts, and interpret results within the study design.

Problem: Researchers need to interrogate evidence after the first analysis

Solution: Evidano's AI chat over documents and analyses allows researchers to ask follow-up questions about themes, excerpts, segments, and source material without repeatedly searching the full corpus.

A research team could use document-based questions to investigate when a biomarker discussion increased, which stakeholder groups discussed treatment access, or how advocacy language changed across a defined period. Outputs should be checked against the underlying source text, particularly in medical and policy research.

Problem: Sensitive health data requires careful handling

Solution: Evidano encrypts data and states that customer data is never used to train third-party models, which is relevant when researchers work with sensitive interviews, patient narratives, or restricted documents.

The PLOS One study demonstrates why privacy safeguards matter: the authors shared tweet IDs rather than full tweet content because of ethical, legal, and platform-policy concerns. Researchers should obtain appropriate approvals, minimize personal data, and avoid treating public availability as equivalent to unrestricted ethical use.

FAQ: Qualitative Social Media Analysis

What is qualitative social media analysis?

Qualitative social media analysis is the systematic interpretation of online language, interactions, and community practices to understand meaning, experience, and social dynamics.

The 2026 PLOS One study applied qualitative analysis to detailed cancer-community conversations after using computational methods to identify relevant topics, actors, and interaction patterns.

How is AI used in qualitative social media research?

AI is used in qualitative social media research to organize large text collections, identify candidate themes, compare segments, surface representative excerpts, and support follow-up questions.

The PLOS One study used BERTopic, transformer-based embeddings, UMAP, and HDBSCAN for computational topic discovery, then had researchers review representative tweets and interpret the conversations qualitatively.

What did the cancer Twitter study analyze?

The cancer Twitter study analyzed more than 8.6 million publications from selected users and examined discussions about precision oncology, biomarkers, targeted therapies, patient advocacy, clinical trials, and healthcare policy.

The researchers studied activity spanning 2010 to 2022 and conducted a focused qualitative analysis of 7, 670 tweets across approximately 3, 000 conversations involving Janet Freeman-Daily.

Can AI qualitative analysis replace human researchers?

AI qualitative analysis should support rather than replace human researchers because interpretation, sampling decisions, ethical judgment, and contextual validation remain necessary.

The PLOS One study illustrates this principle by combining automated topic modeling with manual account classification, review of the 50 most representative tweets per topic, and close reading by three team members.

What ethical issues apply to cancer community social media research?

Cancer community social media research requires attention to privacy, consent, re-identification, platform terms, data retention, and the potential harm of quoting vulnerable users.

The PLOS One authors anonymized most contributors, shared tweet IDs rather than full tweet text, and identified platform changes and deleted posts as constraints on reproducibility. Medical and clinical research analysis is non-diagnostic and should not be used to provide individual treatment advice.

Conclusion and Next Steps for AI Qualitative Research

The 2026 PLOS One study shows that qualitative social media analysis can reveal how patients, advocates, clinicians, and organizations exchange knowledge and influence discussions about precision oncology. The study's combination of 8, 614, 284 publications, stakeholder classification, topic modeling, network analysis, and close reading offers a practical model for analyzing large online communities without losing the detail of lived experience.

AI can reduce the effort required to organize and compare large text collections, but researchers must validate themes, document limitations, protect sensitive data, and preserve human interpretation. Evidano helps research teams move from complex qualitative data to structured themes, cross-segment insights, and evidence-grounded questions. Start your next analysis with Try Evidano for free.

Topics

  • qualitative social media analysis
  • AI qualitative research
  • digital ethnography cancer communities
  • social media analysis for healthcare research
  • AI thematic analysis

Keep reading

Browse all articles