Evidano home
Commentary on News

Qualitative Social Media Analysis With AI: Cancer Communities

Evidano9 min read

Qualitative social media analysis helps researchers understand how people exchange knowledge, build communities, and influence institutions online. The challenge is scale: health researchers may need to examine millions of short, fragmented posts while preserving the context and meaning of individual conversations. A study published by PLOS One on November 20, 2026, shows how mixed methods can address that challenge in online cancer communities. The researchers combined semantic analysis, user classification, retweet networks, and close reading to study precision oncology discussions on Twitter. This article explains what the study found and how AI-enabled qualitative research can help teams move from large social datasets to defensible themes, stakeholder comparisons, and detailed conversation analysis. The discussion is research-focused and non-diagnostic; social media findings should not be treated as medical advice.

Key Takeaways

According to Tweeting with precision: A multiscale longitudinal digital ethnography of online cancer communities, qualitative social media analysis can reveal how patients, advocates, oncologists, and media organizations exchange information and shape discussions about precision oncology.

The PLOS One study analyzed 8, 614, 284 Twitter publications from accounts active in cancer-related communities between January 1, 2010, and December 31, 2022, then used qualitative analysis to explain the social meaning behind large-scale patterns.

  • The PLOS One researchers selected 834 highly active accounts in data collected during March and April 2023; those accounts represented fewer than 3% of relevant users but produced more than 80% of hashtag-related tweets.
  • The PLOS One study identified 116 initial topic clusters and consolidated them into 49 topic groups and 13 overarching categories during analysis of the 2010 to 2022 corpus.
  • The PLOS One study found that precision oncology appeared in 2.4% of tweets, while cancer-related topics represented 50% of the corpus after excluding outliers.
  • The PLOS One study found that oncologists authored 40% of biomarker-related tweets, while advocates and media organizations each authored 17% during the analyzed period.
  • The PLOS One study showed that close reading added context to network and topic patterns, including how ROS1 advocates and oncologists discussed treatment access, clinical trials, and patient organization.

How Qualitative Social Media Analysis Worked in the Cancer Study

The PLOS One study used a mixed-method design that connected computational scale with qualitative interpretation. The researchers began with approximately 30 cancer-related hashtags, identified active users, classified user identities, modeled topics, examined retweet relationships, and then read selected exchanges in detail.

The PLOS One researchers manually classified all 834 selected accounts into eight categories: patients, advocates, oncologists, health professionals, health centers, media, nonprofit organizations, and industries. The researchers based the classification on account descriptions, handles, and consensus discussions about ambiguous cases.

The PLOS One researchers used BERTopic, UMAP, and HDBSCAN to organize the corpus semantically during data processing. The researchers also reviewed the 50 most representative tweets for each topic before assigning broader labels, demonstrating why automated clustering should support rather than replace qualitative judgment.

The qualitative component focused on 7, 670 tweets arranged into approximately 3, 000 conversations involving advocate Janet Freeman-Daily and a wider network of advocates, oncologists, patients, and health professionals. The researchers selected relevant exchanges through team reading and consensus.

The study reported that Twitter discussions combined scientific information, patient support, ordinary socializing, and policy concerns. The researchers described the platform as a space where advocates and oncologists could answer patient questions while also questioning care organization and clinical trial regulation.

One advocate wrote, "Transparency is vital and patients should have a voice at the table! " The quotation illustrates how qualitative coding can capture demands about participation and governance that frequency counts alone may miss.

Findings Snapshot: Themes, Actors, and Time

Date or periodMeasureValueResearch implication
January 1, 2010 to December 31, 2022Full publication corpus8, 614, 284 publicationsA large corpus supports longitudinal and cross-segment analysis.
March to April 2023Selected active accounts834 accountsA focused sample can capture the most active discussion core, but may underrepresent peripheral users.
2010 to 2022Initial topic clusters116 clustersAutomated semantic models can expose a detailed topic structure for later human interpretation.
2010 to 2022Consolidated topic structure49 groups and 13 overarching categoriesHierarchical coding can connect granular themes to higher-level findings.
2010 to 2022Tweets referencing precision oncology2.4%Targeted-topic analysis can isolate a specialized conversation within a broad health corpus.
2010 to 2022Tweets authored by oncologists about biomarkers40%Stakeholder segmentation can show which groups drive specific scientific discussions.
2010 to 2022Tweets authored by advocates and media about biomarkers17% eachComparing stakeholder shares can identify distinct information roles and possible bridges between groups.

Implications for Health and Policy Researchers

Why should researchers combine thematic analysis with interaction analysis?

Researchers should combine thematic analysis with interaction analysis because content shows what communities discuss, while network patterns show how information moves between groups.

The PLOS One study found homophilic retweeting among advocates and oncologists, meaning users tended to share content from their own categories, while advocates also connected more broadly across stakeholder groups. The authors interpreted advocates as potential bridges between medical communities and patients, while noting that the network finding was descriptive.

What can qualitative coding reveal that topic frequency misses?

Qualitative coding can reveal how participants interpret, negotiate, and act on a topic rather than only how often a term appears.

The PLOS One study used the ROS1 case to show how online exchanges preceded the formation of an organized patient group and raised concerns about treatment access, travel costs, clinical trial eligibility, and regulatory policy. Janet Freeman-Daily described Twitter as "the one place where all stakeholders [...] could meet and exchange ideas on something that had life-and-death implications for so many."

What sampling limitation should researchers report?

Researchers should report that a highly active-account sample may overrepresent committed advocates and professionals while underrepresenting casual or less visible participants.

The PLOS One study selected 834 accounts that represented fewer than 3% of the 28, 524 users who used the relevant hashtag communities, although their posts represented more than 80% of hashtag-related tweets. The authors also cautioned that the qualitative focus on one advocate and her network could not represent every advocacy style or patient experience.

How should researchers handle changing social platforms?

Researchers should treat platform changes as part of the research context and document collection dates, access constraints, and reproducibility limits.

The PLOS One study covered Twitter through December 2022 and stated that the platform's later transition to X affected data access, policies, functionality, and algorithms. The authors concluded that the digital era studied "no longer exists, " making archival documentation and transparent methods especially important.

How Evidano Helps With AI Qualitative Social Media Analysis

Problem: Large volumes of fragmented text make manual synthesis slow

Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze social posts, interview transcripts, reports, papers, and other text-based evidence.

Evidano can support thematic, content, frequency, and cross-segment analyses so research teams can move from a large corpus to recurring themes, stakeholder differences, and interpretable findings. Teams can explore the platform's capabilities through Evidano's features.

Problem: Stakeholder groups may discuss the same issue differently

Solution: Researchers can use cross-segment analysis to compare themes across groups such as patients, advocates, clinicians, media organizations, or geographic and demographic segments.

The PLOS One study demonstrates the value of this approach by comparing how oncologists, advocates, patients, and media participants contributed to biomarker discussions between 2010 and 2022.

Problem: High-level themes can hide important conversations

Solution: Researchers can combine hierarchical codes and subcodes with AI chat over source documents and analyses to investigate representative excerpts and unexpected patterns.

The PLOS One study moved from 116 initial clusters to 49 groups and 13 overarching categories, then used close reading to interpret specific ROS1 conversations. A similar workflow can preserve both corpus-level structure and conversation-level detail.

Problem: Sensitive health conversations require careful data handling

Solution: Research teams can use a platform with encryption and a policy that data is never used to train third-party models, while still applying their own consent, anonymization, and governance procedures.

The PLOS One study withheld full tweet content and shared tweet IDs because of ethical, legal, and platform-policy concerns. Researchers should treat AI analysis as part of a broader research governance process, not as a substitute for ethical review.

FAQ: Qualitative Social Media Analysis

What is qualitative social media analysis?

Qualitative social media analysis is the systematic interpretation of posts, conversations, interactions, and participant roles to understand meaning and social processes.

The PLOS One study combined semantic analysis, user classification, retweet networks, and close reading to examine how online cancer communities discussed precision oncology and patient advocacy.

How can AI support qualitative social media research?

AI can support qualitative social media research by organizing large text collections, identifying candidate themes, comparing segments, surfacing representative excerpts, and enabling researchers to ask questions across a corpus.

The PLOS One study used computational topic modeling and network analysis before researchers reviewed representative tweets and selected conversations qualitatively. The study supports a human-guided model in which AI expands coverage while researchers validate meaning and context.

How many tweets did the cancer community study analyze?

The PLOS One study analyzed 8, 614, 284 publications from selected cancer-community accounts, with the account histories covering posts published between January 1, 2010, and December 31, 2022.

The corpus included 4, 389, 619 tweets and 3, 031, 281 retweets according to the study's methods section, with the remaining publications belonging to other publication types counted in the dataset.

What did the study find about patient advocates?

The PLOS One study found that patient advocates contributed to information exchange across stakeholder groups and often addressed scientific, access, policy, and patient-support issues.

The researchers found that advocates were proportionally active in biomarker discussions and appeared to bridge patients and medical professionals, although the network evidence was descriptive and did not prove causal influence.

Can social media research provide medical advice?

Social media research cannot replace clinical guidance, diagnosis, or individualized medical advice.

The PLOS One study analyzed public discourse about cancer care, targeted therapies, and clinical trials as a research object. Researchers and readers should verify medical claims through qualified clinicians and authoritative clinical sources.

Conclusion and Next Steps

The PLOS One study shows that qualitative social media analysis is strongest when computational scale and human interpretation work together. The study's 8, 614, 284-publication corpus revealed broad patterns, while close reading explained how advocates and oncologists discussed treatment access, research, and patient organization.

AI-enabled qualitative research can make this workflow more manageable by supporting thematic coding, cross-segment comparisons, frequency analysis, and focused exploration of source excerpts. Researchers should still document sampling decisions, validate AI-generated themes, protect sensitive data, and report platform limitations.

Teams that want to analyze interviews, social posts, reports, or open-ended survey responses can try Evidano for free.

Topics

  • qualitative social media analysis
  • AI qualitative research
  • cancer community analysis
  • digital ethnography
  • social media thematic analysis

Keep reading

Browse all articles