Researchers, UX teams, and policy analysts evaluating digital health in fragile settings need repeatable, rapid ways to turn interviews and app logs into operational recommendations. This post refracts the 6-week mPallCare feasibility study (published June 26, 2026) through the lens of qualitative research workflows and shows how to run a reproducible qualitative analysis of digital health in humanitarian settings using AI tools like Evidano. You will get a quick summary of what the study found, a snapshot of key metrics, and a compact 7-step workflow you can run on your own transcripts, dashboards, and survey exports.
Key Takeaways
Evidano is an AI-powered qualitative data analysis platform that helps teams convert interview transcripts, app logs, and outcome scales into actionable themes and prioritized product and policy fixes.
The mPallCare pilot shows structured symptom and outcome monitoring is feasible in a refugee settlement, but translation, synchronization, and sustaining engagement are the main friction points.
- The mPallCare pilot enrolled 32 participants, ran a 6-week follow-up, and the full paper was published in JMIR mHealth and uHealth on June 26, 2026 (JMIR mHealth and uHealth).
- Combined completion of scheduled reports was 429/640 (67.0%), exceeding the predefined feasibility threshold of 65%; symptom reports were 163/192 (84.9%) and outcome reports were 266/448 (59.4%).
- Reproducible analysis in one week is possible by ingesting transcripts, normalizing language with local dictionaries, running AI-assisted coding, quantifying themes, and producing stakeholder-ready visuals using the 7-step workflow.
Fast take: what the paper tested and why it matters
The paper tested the feasibility and acceptability of mPallCare, a mobile app plus clinician dashboard for palliative care symptom and outcome monitoring, in Bidibidi Refugee Settlement, Uganda. The pilot enrolled 32 people with advanced cancer and ran scheduled reporting over 6 weeks, full paper at JMIR mHealth and uHealth.
- Why analysts care: routine, structured patient-reported data in displacement settings can inform service decisions, and qualitative data from users is essential to interpret uptake, translation needs, and workflow friction.
- Payoff in this post: concrete, reproducible steps to convert the study’s interview transcripts, app logs, and POS scores into themes, segment comparisons, and prioritized product fixes using AI-assisted qualitative workflows.
Findings snapshot
| Date / Metric | Value | Source / Note |
|---|---|---|
| Publication | June 26, 2026 | JMIR mHealth and uHealth |
| Study period (enrolment) | Active recruitment Aug–Oct 2022; 6-week follow-up per participant | Paper methods |
| Location | Bidibidi Refugee Settlement, Uganda | Largest refugee settlement in Uganda |
| Participants | N=32 people living with advanced cancer; 20 VHTs trained | Demographics in Table 1 |
| Symptom reports | 163/192 (84.9%) completed | Weekly symptom form (6 timepoints) |
| Outcome reports | 266/448 (59.4%) completed | Outcome every 3 days (≈14 timepoints) |
| Combined completion | 429/640 (67.0%) of scheduled reports | Feasibility threshold ≥65% |
| Top reported symptoms | Headache (27/32, 84.4%); muscle pain (27/32, 84.4%); dizziness (26/32, 81.3%) | Table 2 |
| Languages available (initial) | English, French, Swahili, Juba Arabic | Local language expansion planned |
What happened: methods and core results (plain English)
The methods combined VHT-assisted mobile reporting and a clinician dashboard to capture a 17-item symptom checklist and the 7-item APCA Palliative Outcome Scale (POS). VHTs completed symptom reports weekly and outcome reports every 3 days, and clinicians could flag and respond from a web dashboard.
- Feasibility: the combined completion rate of 67% met the predefined feasibility threshold of at least 65%, driven by high symptom-report completion (84.9%) and lower-frequency outcome reporting (59.4%).
- Acceptability: interviews with 10 patients and 10 clinical/VHT staff reported improved communication, quicker medication access, and enhanced patient empowerment, while also identifying navigation, translation, synchronization, and device-availability issues.
- Engagement trend: submissions declined modestly across the 6 weeks, attributed to technical sync errors, VHT workload, and disease progression reducing participant energy.
Implications for researchers, UX teams and policy analysts
For qualitative researchers
For qualitative researchers, mixed methods pilots in fragile settings must pair descriptive metrics with rapid thematic analysis to explain behavior. Mixed methods pilots should pair completion rates and symptom frequencies with rapid thematic coding to explain why outcome reporting lagged while symptom reporting remained high.
Prioritize reproducible codebooks and cross-segment comparisons (by age, mobile access, caregiver presence) so that qualitative themes can be quantified and presented to stakeholders.
For UX / product teams
For UX and product teams, translation and microcopy are integral UX features that materially affect validity. Test translation with real users and store local dictionary mappings.
Design for low-energy users: reduce frequency or length of forms or provide audio and pictorial inputs; ensure device access plans (clinic storage versus home) and offline sync resilience.
For health policy & program teams
For health policy and program teams, structured patient-reported outcome data can inform district planning and integration with DHIS2, provided systems capture action metadata like referral and psychosocial linkages. Ensure escalation pathways and partner services before scaling routine symptom surveillance to respect ethical obligations.
Do more, faster with Evidano (mapped to this use case)
Ingest: transcripts, app logs, and POS spreadsheets
Evidano ingests interview audio and transcripts, clinician dashboard exports, and Excel/CSV survey exports so teams can analyze the full mixed-methods corpus in one place. Use a custom import to combine VHT notes, interview transcripts (language variants), and time-stamped symptom and outcome logs for longitudinal analysis.
Automated thematic + frequency analysis
Evidano runs AI-assisted thematic coding across interviews to surface acceptability themes such as usability, translation, and sync errors, and quantifies theme prevalence across segments like age and mobile access. Evidano provides theme frequencies, co-occurrence matrices, and representative quotes to help prioritize product fixes that affect the most users.
Translation and transcription with local dictionaries
Evidano supports uploading local vocabulary to improve translation accuracy and normalizes terms across transcripts so equivalent expressions map to the same code. Transcription includes PII redaction options to satisfy ethical constraints in humanitarian research.
Visualize patterns and cross-segment differences
Evidano generates co-occurrence networks (symptoms linked to concerns), hierarchical code trees, and time-series plots of POS domain scores to show trajectories like those in the mPallCare pilot. Export stakeholder-ready visuals and a concise findings deck for Ministry of Health or funder briefings.
Secure, auditable, and research-grade
Evidano keeps data encrypted and does not use customer data to train third-party models, and Evidano maintains an audit trail of coding decisions for ethics committees or ministry reviews.
7-step workflow: reproduce the mPallCare qualitative analysis in ≈1 week
The 7-step workflow reproduces the mPallCare qualitative analysis in approximately one week, assuming access to interview audio/transcripts, VHT app logs, and POS and roster spreadsheets.
- 1) Ingest: Upload raw transcripts, audio, and CSV exports into Evidano, and attach a codebook or start with auto-coding.
- 2) Normalize language: Load a custom dictionary for local terms (Juba Arabic, Bari) and run batch translation to a single analysis language.
- 3) Auto-code & refine: Run AI thematic coding, then review and adjust the generated code hierarchy (themes to subthemes).
- 4) Quantify: Produce frequency tables for themes, symptom mentions, and completion rates (aligns with the paper’s 84.9% symptom and 59.4% outcome rates).
- 5) Cross-segment analysis: Compare themes by mobile access (yes/no), caregiver presence, and week to identify drop-off drivers.
- 6) Visualize: Export co-occurrence networks (for example translation issues linked with navigation friction) and time-series of POS domains.
- 7) Share & iterate: Generate a one-page executive brief and an interactive dashboard for clinical teams, and schedule targeted follow-up interviews guided by AI-identified gaps.
FAQ: AI-enabled qualitative analysis
Can AI reliably surface translation-related usability issues?
+
Yes, AI can reliably surface translation-related usability issues when combined with custom dictionary normalization and co-occurrence analysis. By normalizing local vocabulary and identifying clusters where participants mention not understanding or repeat paraphrases, AI highlights microcopy failures that require targeted fixes.
How do you compare segments (for example, those with versus without phone access)?
+
Tag records on import with segment labels and run cross-segment frequency and sentiment comparisons to identify differential barriers. Tagging on fields such as mobile_access (yes/no) and caregiver presence lets teams prioritize interventions based on differential theme prevalence.
Is this approach ethical with sensitive health data?
+
Yes, the approach can be ethical when it uses PII redaction, encrypted storage, and restricted access, and follows the approved ethics protocol. Evidano does not use customer data to train third-party models, and teams should always align analysis steps with their approved ethics procedures.
Wrapping up: next steps
The mPallCare pilot shows that structured symptom and outcome monitoring is feasible in a refugee settlement, and that translation, sync reliability, and sustaining engagement are the main friction points. Turning those qualitative signals into prioritized product and policy fixes is a repeatable workflow.
- If you want to replicate this analysis on the mPallCare corpus or your own interviews and app logs, start a pilot import into Evidano and run the 7-step workflow above.
- Ready to move from insight to action? Try Evidano for free to run a guided pilot with your transcripts and dashboards and get a prioritized findings brief you can share with clinical teams and funders.
Topics
- qualitative analysis
- digital health humanitarian
- mHealth palliative care
- AI-assisted qualitative coding
Keep reading
- Commentary on NewsBeyond Speed: AI Qualitative Data AnalysisLearn how AI qualitative data analysis can accelerate coding without replacing researcher judgment, with practical safeguards for traceable, ethical interpretation.
- Commentary on NewsAI-assisted qualitative coding: Qualcoder MCP serverExplainer of the Qualcoder MCP server for AI-assisted qualitative coding, including the PyPI release date, privacy flow to Anthropic, safety steps for researchers, and practical next steps.
- Commentary on NewsQualitative Analysis: SLP Job Readiness (2026)PLOS One July 21, 2026 qualitative study (n=15, 5 focus groups) on SLP job readiness, plus an AI-enabled workflow for rapid transcription, coding, and curriculum fixes.