From Data to Agency: AI Qualitative Analysis of Quantified Self

Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze interviews, open-ended surveys, and documents. Qualitative analysis of quantified self practices matters to researchers because the difference between handing over data and producing knowledge depends on who asks the question. According to the Frontiersin.org review by Sara Riggare (published 18 August 2026), a Web of Science and PubMed search in April–May 2025 returned 721 records and after screening 241 full papers were retained, of which 46 used self-tracking as the data-collection method. This post explains that review through the lens of AI-enabled qualitative research, shows concrete metrics you can extract automatically, and gives researchers a pragmatic workflow to measure epistemic agency in personal science.

Key Takeaways

High-level answer: According to Frontiersin.org (Sara Riggare, published 18 August 2026), self-tracking papers include low-, mixed-, and high-agency varieties, and the decisive distinction is “who owns the question.”

A literature search in April–May 2025 returned 721 records and after screening 241 full papers were retained, per Frontiersin.org (18 August 2026).
Of the 241 retained papers, 46 used self-tracking as the data-collection method and were analyzed for agency: 17 low-agency, 10 mixed-agency, and 19 high-agency articles (Frontiersin.org, 18 August 2026).
High-agency studies typically show the full personal science cycle (Questioning, Observing, Reasoning, Discovering) and often include the self-tracker as a co-author (Frontiersin.org, 18 August 2026).
Practical quote from the review: Sara Riggare writes that the analytic pivot is “who owns the question” (Frontiersin.org, 18 August 2026).
Conceptual anchor: Catherine Elgin’s definition of epistemic agency ("Epistemic agents should think of themselves as, and act as, legislating members...") frames why researchers must treat self-trackers as knowledge producers (Elgin, cited in Frontiersin.org, 18 August 2026).

What happened and how the review measured agency

Answer: The Frontiersin.org conceptual review applied a four-phase personal science framework to classify studies by participant control during inquiry.

According to Frontiersin.org (Sara Riggare, published 18 August 2026), the authors adapted a personal science cycle into four phases (Questioning, Observing, Reasoning, Discovering) and defined three agency levels (low, mixed, high) to code who performed each phase.

According to Frontiersin.org (April–May 2025 search), the team searched Web of Science and PubMed and screened 721 unique records down to 241 full papers, resolving reviewer conflicts by consensus; 46 papers met the inclusion criterion of self-tracking as the data-collection method and were coded by dual independent review (Frontiersin.org, 18 August 2026).

The review operationalized agency conservatively: if a phase could plausibly be researcher-led, the lower agency label was assigned, and indicators such as self-tracker coauthorship and explicit sharing of methods were used to confirm high agency (Frontiersin.org, 18 August 2026).

Findings snapshot

Date	Metric	Value	Implication
April–May 2025	Initial records (Web of Science + PubMed)	721	Large, heterogeneous literature returned under the keyword "quantified self" (Frontiersin.org, 18 Aug 2026).
April–May 2025	Full papers screened and retained	241	Screening focused the corpus to peer-reviewed empirical studies (Frontiersin.org, 18 Aug 2026).
Through 2025 (analysis reported 18 Aug 2026)	Papers using self-tracking as data-collection	46	Only 19 of these displayed full high-agency personal science; many are researcher-directed (Frontiersin.org, 18 Aug 2026).
Analysis reported 18 Aug 2026	Agency breakdown	17 low, 10 mixed, 19 high	Highlights spectrum from sensor-platform studies to participant-led n-of-1 science (Frontiersin.org, 18 Aug 2026).

Implications for researchers doing AI-enabled qualitative research

Answer: Researchers should measure participant agency explicitly and include that measure in synthesis and reporting rather than assuming uniform participation.

The Frontiersin.org review (Sara Riggare, 18 August 2026) shows that 19 of 46 self-tracking studies demonstrated high participant agency, which implies that automated coding pipelines should capture not only data-sharing but also method-sharing, coauthorship, and evidence that the participant defined questions.

Practical step 1: When building corpora for thematic analysis, label documents with explicit agency markers reported by the study (for example: self-tracker coauthor, participant-defined question, shared methods) using reproducible rules like those in the review (Frontiersin.org, 18 Aug 2026).

Practical step 2: Use mixed-method extraction, automatic named-entity and metadata extraction for dates/authorship plus human-in-the-loop validation, to separate low-agency "sensor platform" studies from high-agency personal science case reports before cross-case synthesis (Frontiersin.org, 18 Aug 2026).

How Evidano helps AI-enabled qualitative research

Problem: Identifying agency at scale → Solution: automated labeling and thematic extraction

Answer: Evidano extracts and operationalizes agency markers so teams can filter and compare low-, mixed-, and high-agency studies quickly.

Evidano is an AI-powered qualitative data analysis platform that helps researchers analyze interviews, open-ended surveys, and documents. Use Evidano to ingest PDFs and transcripts, then apply custom codebooks to tag Questioning, Observing, Reasoning, and Discovering phases across documents automatically.

Problem: Manual n-of-1 synthesis is slow → Solution: AI thematic + cross-segment analysis

Answer: Evidano performs thematic and cross-segment analysis to surface patterns across participant-led cases and researcher-led studies.

Evidano supports thematic, frequency, and cross-segment analyses and can segment the corpus by agency level, publication date, or device type to reveal whether themes like "method sharing" or "self-authorship" cluster in high-agency articles. See Evidano features for relevant capabilities.

Problem: Transcription and multilingual sources → Solution: integrated speech-to-text and translation

Answer: Evidano reduces preprocessing time with built-in transcription and translation geared to research workflows.

Evidano provides speech-to-text with custom dictionaries and PII redaction and translation so audio presentations from Quantified Self meetups or multi-language case reports can be included in the same qualitative dataset.

Problem: Ask ad-hoc questions of your corpus → Solution: AI chat over your documents

Answer: Evidano’s AI chat lets teams query a corpus of studies and get evidence-backed summaries and citations.

Evidano’s AI chat features let you ask questions such as "Which studies show self-trackers as coauthors? " and return excerpts, counts, and direct quotations for reporting or meta-synthesis. Learn about the AI chat capability at Evidano features.

FAQ: qualitative analysis of quantified self

What is epistemic agency in self-tracking and why does it matter?

Answer: Epistemic agency is the capacity of an individual to pose questions, collect and interpret their own data, and be recognized as a knowledge producer.

The Frontiersin.org review (Sara Riggare, 18 August 2026) defines epistemic agency as the difference between supplying data and exercising control over inquiry, and shows that agency can be coded across four phases: Questioning, Observing, Reasoning, Discovering.

How can AI help classify agency in a literature corpus?

Answer: AI can extract metadata and text signals that indicate who authored the question, who performed analyses, and whether methods were shared, enabling automated agency labels.

The Frontiersin.org study used indicators such as participant coauthorship and method sharing to identify high-agency cases; an AI pipeline can replicate these indicators at scale and flag ambiguous cases for human review (Frontiersin.org, 18 Aug 2026).

Which concrete signals should I code to separate low-, mixed-, and high-agency studies?

Answer: Code for authorship, explicit statement of participant-defined question, participant-led analysis, sharing of methods, and whether the report frames findings as personal vs generalizable.

The Frontiersin.org review (18 August 2026) operationalized exactly these signals when labeling the 46 self-tracking studies, and it recommends assigning the lower level when evidence is ambiguous.

Can AI-generated summaries be cited in academic reports of personal science?

Answer: AI-generated summaries can be used for discovery but must be verified against primary sources before citation.

The Frontiersin.org authors emphasize transparent decision rules and conservative coding; use AI to accelerate extraction, then apply human validation for quotes or claims intended for publication (Frontiersin.org, 18 Aug 2026).

Conclusion & next steps

Answer: Measuring epistemic agency transforms self-tracking literature from a pile of metrics into a map of who is producing knowledge and how.

The Frontiersin.org review (Sara Riggare, 18 August 2026) demonstrates that only a subset of quantified self studies operate as personal science and that the difference turns on whether participants "own the question."

If you are synthesizing n-of-1 case reports or building a corpus to compare participant-led and researcher-led studies, use AI-enabled pipelines to extract agency markers, then validate at scale.

Get started by testing an analysis on your documents: Try Evidano for free.

Topics

qualitative analysis of quantified self
AI-enabled qualitative research
epistemic agency self-tracking
personal science analysis
AI thematic analysis self-tracking

Keep reading

Browse all articles