AI Qualitative Analysis in Healthcare: PROMIS CAT Lessons

According to the Feasibility of patient-reported outcomes measurement information system (PROMIS) computerized adaptive tests in patients receiving physical therapy: A mixed-methods study, published in PLOS One on November 13, 2026, PROMIS CATs were feasible and generally acceptable in primary care physical therapy, but adoption depended on reducing duplicate questionnaires, improving score interpretation, and fitting the technology into clinical workflows.

AI-enabled qualitative analysis can help healthcare researchers reach similar implementation insights faster by linking interview themes, survey percentages, participant groups, and contradictory findings in one auditable analysis.

• The 2026 PLOS One study covered 14 physical therapy practices, with 25 therapists completing a survey and 10 therapists plus 9 patients participating in interviews.
• In the survey conducted from May 2 to June 20, 2022, 80% of therapists expressed interest in using PROMIS CATs, while only 28% reported regular daily-practice use.
• The 2026 study found that 100% of surveyed therapists considered PROMIS CATs easy to administer, but only 40% believed CATs took little time.
• Patients generally described PROMIS CATs as quick and relevant, while therapists identified portal problems, uncertain score interpretation, and simultaneous legacy PROMs as barriers.
• The study’s qualitative analysis used hybrid inductive and deductive thematic analysis, with two researchers independently coding transcripts before consensus meetings.

The 2026 PLOS One study evaluated whether Dutch-Flemish PROMIS CATs could work in routine primary care physical therapy for adults with musculoskeletal conditions affecting the spine and upper extremities.

The researchers used a concurrent mixed-methods design. According to the study methods, a survey assessed therapists’ experiences across acceptability, demand, practicality, and implementation, while semi-structured interviews explored those same domains with therapists and patients.

The study’s interview transcripts were analyzed through hybrid thematic analysis. Two researchers independently performed open coding, compared codes during consensus meetings, and used a third researcher to resolve disagreements when necessary.

PROMIS CATs selected questions dynamically from item banks based on previous answers. According to the study protocol, each CAT administered at least 4 and at most 12 items, stopping when the estimated standard error reached a threshold corresponding to approximately 0.95 reliability.

The study collected interviews between March 21 and November 23, 2022. Interviews lasted between 32 and 62 minutes, with a mean duration of 47 minutes, and transcripts were checked for accuracy after automatic or professional transcription.

AI qualitative analysis in healthcare is most useful when it preserves the difference between what participants said, how often a view appeared, and what researchers infer from the evidence.

The 2026 PLOS One study demonstrates why cross-segment analysis matters. Only 24% of surveyed therapists believed patients were satisfied with PROMIS CATs, yet interviewed patients generally described the CATs as quick, simple, and relevant. An AI-assisted workflow can surface this divergence by comparing therapist survey data with patient interview codes instead of collapsing both groups into one sentiment score.

The study also shows why thematic analysis must connect benefits to implementation conditions. One therapist said, "If the CAT can replace NPRS, PSFS, and Quebec while giving the same information, that’s a win-win" (PT#6). The quotation supports a replacement strategy, but the same study found that legacy PROMs were still administered alongside CATs, increasing respondent burden.

AI can support this work by grouping references to questionnaire fatigue, portal access, score interpretation, item relevance, and clinical reasoning into a structured code hierarchy. Researchers must still review quotations, define the analytic framework, and determine whether a theme represents a convergence, complementarity, or divergence between data sources.

The study’s findings are research evidence, not clinical guidance. Healthcare teams should use qualitative findings to inform implementation and patient-centered discussion, not to diagnose individual patients.

The 2026 PLOS One study shows that PROMIS CATs can support patient-centered conversations, but successful implementation requires more than a technically functional questionnaire. Qualitative analysis in healthcare must connect participant quotations and themes to survey frequencies, subgroup differences, and real-world workflow constraints.

AI-enabled analysis can help research teams manage this complexity by organizing transcripts and spreadsheets, comparing patient and clinician perspectives, and making supporting evidence easier to retrieve. Researchers should use AI to accelerate analysis while retaining human oversight of coding quality, privacy, interpretation, and clinical context.

Healthcare teams that want a structured way to analyze interviews, open-ended survey responses, and implementation documents can Try Evidano for free.