Evidano home
Commentary on News

From 8.6M Tweets to Insight: AI Qualitative Analysis

Evidano9 min read

Cancer researchers, patient advocacy teams, and health policy analysts increasingly need to understand not only what online communities discuss, but also how patients, clinicians, media, and organizations influence one another. The challenge is scale: a decade of social posts can contain millions of messages, while the most important evidence may sit inside a small number of conversations. AI qualitative analysis of social media helps researchers connect large-scale patterns with the context, language, and interaction dynamics that give those patterns meaning. The November 2026 study "Tweeting with precision" provides a useful example. The researchers combined semantic analysis, user classification, retweet networks, and close reading to examine cancer communities on Twitter between 2011 and 2022. This article explains the study's findings, shows why mixed-method analysis matters, and translates the workflow into practical guidance for teams conducting qualitative social media analysis.

Key Takeaways

According to the PLoS One study, "Tweeting with precision", AI-assisted qualitative analysis of social media can connect millions of posts with detailed evidence about how patients, advocates, oncologists, and other stakeholders exchange knowledge.

The study found that cancer communities on Twitter were not only information channels: according to the November 2026 research, patient advocates helped bridge clinical expertise, patient experience, treatment access concerns, and policy discussions.

  • The November 2026 PLoS One study analyzed 8, 614, 284 publications from accounts active in cancer-related Twitter communities between 2011 and 2022.
  • The researchers selected 834 highly active accounts, fewer than 3% of the 28, 524 users identified, although those accounts produced more than 80% of hashtag-related tweets.
  • The study's topic model initially identified 116 clusters, which the researchers consolidated into 49 topic groups and 13 overarching categories.
  • According to the study, 2.4% of tweets referred to precision oncology, while cancer-related topics represented 50% of the corpus after outliers were excluded.
  • The researchers' qualitative analysis examined 7, 670 tweets organized into approximately 3, 000 conversations involving Janet Freeman-Daily and a wider network of advocates, oncologists, patients, and health professionals.

What the Cancer Social Media Study Measured

The November 2026 PLoS One study used a mixed-method digital ethnography to examine both the structure of cancer-related Twitter conversations and the meaning of selected exchanges.

The researchers began with approximately 30 cancer-related hashtags and then used an account-based strategy to identify highly active users. The research team classified 834 accounts into eight stakeholder categories: patients, advocates, oncologists, health professionals, health centers, media, nonprofit organizations, and industries.

For large-scale semantic analysis, the researchers used BERTopic with transformer-based document embeddings, UMAP for dimensionality reduction, and HDBSCAN for clustering. The researchers then reviewed the 50 most representative tweets for each topic to validate and label the resulting themes.

The researchers also analyzed retweet networks to study homophily, meaning the tendency of users to interact with others who share similar identities or roles. The study reported that oncologists and advocates tended to share content within their own groups, while advocates also connected information from a wider range of stakeholders.

The qualitative component added context that frequency counts alone could not provide. The researchers found that conversations included treatment information, clinical trial access, side effects, patient support, healthcare policy, and ordinary social interaction.

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, " according to the study's cited source. The researchers also quoted the #LCSM organizers, who said the chats "brought together a variety of stakeholders in the lung cancer community on equal footing and removed silos to communications."

Findings Snapshot: AI Qualitative Analysis of Social Media

Date or periodMetricValueResearch implication
January 2011 to December 2022Longitudinal coverageMore than 10 yearsAccording to the November 2026 PLoS One study, the long time span captured changes in communities, biomarkers, therapies, and advocacy.
Collection completed April 30, 2023Publications in the main dataset8, 614, 284The study demonstrated why automated semantic and network analysis is useful for screening very large social datasets.
2023 account selectionHighly active accounts834 of 28, 524 usersThe researchers focused on the community's highly engaged core, which may overrepresent professional and committed voices.
November 2026 studyTopic structure116 clusters, 49 groups, 13 categoriesThe study combined machine-generated clusters with human interpretation to create a usable thematic structure.
2011 to 2022 corpusPrecision oncology content2.4% of tweetsThe study showed that a relatively narrow biomedical topic can be embedded within broader social, clinical, and policy conversations.
2016 to 2018ROS1 advocacy activityAdvocates outpaced oncologists in tweet volumeThe study linked this activity to the development of ROS1-focused patient advocacy and treatment-access discussions.

Implications for Cancer Researchers and Patient Advocacy Teams

Cancer researchers should treat qualitative social media analysis as a layered process that combines prevalence, context, stakeholder identity, and interaction patterns.

The November 2026 PLoS One study showed why topic frequency is not enough. The study found that 25% of labeled tweets concerned mundane subjects, demonstrating that ordinary socializing helped maintain the relationships through which information circulated.

Patient advocacy teams can use cross-segment analysis to compare how patients, advocates, clinicians, media, and organizations frame the same biomarker or treatment. The study reported that oncologists authored 40% of biomarker-related tweets, while advocates and media each authored 17%, but advocates were proportionally more active because they represented a smaller account group.

Health policy analysts can use qualitative coding to identify issues that do not appear as formal policy documents. The study's close reading found discussions about clinical trial eligibility, compassionate use, off-label treatment, testing, access barriers, and patient participation in research.

Researchers should document sampling limits before generalizing from social media. The November 2026 study noted that its account-based sample emphasized highly active users, focused largely on an Anglo-American context, and reflected a Twitter environment that changed substantially after the study period.

Medical and clinical research teams should use social media analysis for research and community understanding, not diagnosis or individualized treatment advice. The November 2026 PLoS One study is a non-diagnostic analysis of online discourse and stakeholder interaction.

How Evidano Helps with Qualitative Social Media Analysis

Problem: Millions of posts make manual screening impractical

Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze social media exports, interview transcripts, survey responses, papers, reports, and other documents.

Evidano can help teams organize large text collections for thematic, content, frequency, and cross-segment analysis. Researchers can use these analyses to move from broad patterns to specific examples that require human interpretation.

Teams can learn more about Evidano's research workflow on the Evidano features page.

Problem: Stakeholder differences are hidden in aggregate results

Solution: Researchers can compare themes across stakeholder or demographic segments, such as patients, advocates, clinicians, media, and nonprofit organizations, when those categories are represented in the source data.

The November 2026 PLoS One study used eight manually assigned stakeholder categories. A comparable cross-segment workflow can help researchers examine who raises an issue, who responds, and how language differs between groups.

Problem: Important meaning is lost in frequency counts

Solution: Researchers can combine frequency analysis with thematic coding and AI chat over the source documents. Evidence remains tied to the underlying text rather than being reduced to a word count.

The November 2026 PLoS One study showed the value of this combination by pairing topic modeling with close reading of representative tweets and selected conversations.

Problem: Complex themes need transparent exploration

Solution: Evidano provides visualizations such as word clouds, co-occurrence networks, and hierarchical codes with subcodes, helping research teams inspect relationships among concepts.

Evidano also provides encrypted data handling and states that customer data is never used to train third-party models. Teams conducting health-related research should still obtain appropriate permissions, remove unnecessary personal information, and follow applicable ethics and platform requirements.

FAQ: AI Qualitative Analysis of Social Media

What is AI qualitative analysis of social media?

AI qualitative analysis of social media uses artificial intelligence to identify themes, patterns, segments, and relationships in text while preserving the context needed for human interpretation.

The November 2026 PLoS One study illustrates the approach by combining semantic modeling, stakeholder classification, retweet-network analysis, and manual reading of selected conversations.

How did the cancer study analyze more than 8 million publications?

The November 2026 PLoS One study used a staged workflow that combined hashtag and account selection, automated topic modeling, stakeholder classification, retweet-network analysis, and qualitative review.

The researchers analyzed 8, 614, 284 publications from the complete tweet histories of 834 highly active accounts and then examined selected conversations in greater depth.

What did the study find about patient advocates?

The November 2026 PLoS One study found that patient advocates acted as bridges between patients, oncologists, media, and other organizations in cancer-related online discussions.

The study reported that advocates discussed a wider range of issues than oncologists and often connected biomedical information with treatment access, clinical research, policy, and patient experience.

Why combine thematic analysis with network analysis?

Thematic analysis shows what people discuss, while network analysis shows how information and attention move between stakeholder groups.

The November 2026 PLoS One study used both methods because topic frequencies alone could not explain whether patients, advocates, oncologists, or media were shaping the circulation of information.

What are the main limitations of this type of research?

The main limitations are sampling bias, platform change, privacy constraints, language and regional bias, and the risk of overrepresenting highly active users.

The November 2026 PLoS One study acknowledged that its 834-account sample represented fewer than 3% of identified users, that its qualitative analysis centered on one advocate's network, and that Twitter's transition to X limited future reproducibility.

Conclusion and Next Steps

The November 2026 PLoS One study demonstrates that meaningful qualitative insight can emerge when large-scale semantic analysis is combined with stakeholder classification, network analysis, and close reading.

For cancer researchers and patient advocacy teams, the practical lesson is to analyze both the visible themes and the relationships that make knowledge circulate. AI can accelerate discovery across millions of posts, while researchers remain responsible for sampling decisions, interpretation, privacy, and ethical use.

Evidano gives research teams a way to organize documents and text data, compare segments, investigate themes, and ask questions across their evidence base. Try Evidano for free to turn complex qualitative datasets into structured research insights.

Topics

  • AI qualitative analysis of social media
  • qualitative social media analysis
  • AI thematic analysis
  • digital ethnography cancer research
  • patient advocacy research

Keep reading

Browse all articles