Effectiveness of Generative Artificial Intelligence-Enabled Teaching on Clinical Reasoning in Healthcare Education: A Systematic Review and Meta-Analysis Protocol. (Preprint)

Background:
Clinical reasoning skills are key competencies required by healthcare professionals. Generative Artificial Intelligence (GAI)-enabled teaching may be able to offer personalised learning at scale, however its effectiveness compared with traditional teaching methods for developing clinical reasoning capabilities remains underexplored.

Objective:
Aim: Evaluate the effectiveness of GAI-enabled teaching methods compared to traditional teaching methods on clinical reasoning outcomes in healthcare education. Primary objective: To synthesise and compare evidence of effectiveness on clinical reasoning capabilities between GAI-enabled and traditional teaching methods in healthcare education.

Methods:
The study has been registered with PROSPERO, registration number CRD420261284944, and will adhere to the preferred reporting for systematic review and meta-analysis protocol guidelines (PRISMA-P). Study eligibility: Population: Healthcare professionals–undergraduate / postgraduate and post-registration. Intervention: GAI-enabled teaching methods. Comparator: Traditional teaching methods. Outcome: Educational effectiveness in developing clinical reasoning capabilities. Study settings: University, higher education, clinical settings. Publication status: No limitations on language or geography. Study design: Randomised controlled trials (RCTs), non-randomised controlled trials (non-RCTs), quasi-experimental studies and cohort studies with a suitable control arm will be included. Search strategy: Database search from 2022, including Embase, Medline, CINAHL, Scopus and CENTRAL, supplemented by a Google Scholar search, reference list and citation search of eligible studies. Data management / extraction: Covidence systematic review software will be used for de-duplication and data screening . Title / abstract and full text screening will be undertaken by two independent reviewers, arbitrated by a third reviewer if no consensus is reached. Data extraction will be informed by reporting frameworks for AI interventions and clinical reasoning assessment and the following guidelines: CONSORT for RCTs, TREND for non-RCTs / quasi experimental studies and STROBE for cohort studies. Data synthesis: We will use SPSS software to combine and calculate overall effect. If data shows moderate heterogeneity, (I2 >50% or P<0.1) we will consider completion of a meta-analysis using a random effects model. If heterogeneity exceeds I2>80% we will describe the data using a narrative synthesis. Sensitivity / subgroup analysis: Sensitivity analysis will be used to explore possible sources of heterogeneity. A subgroup analysis may include type of GAI used and/or healthcare professional background. Risk of bias. ROB -2 tool will be used for RCTs, ROBINS-IV2 tool will be used for non-RCTs, quasi-experimental studies and ROBINS-E tool for interventional cohort studies by two independent reviewers, arbitrated by a third reviewer as required. Certainty of evidence: GRADE will be used by 2 independent reviewers to assess certainty of evidence. Ethical approval: Not required.

Results:
Results Expected data collection - Feb / Mar 2026

Conclusions:
We anticipate that the findings will provide evidence-informed insight into the potential role of GAI technologies in supporting the development of clinical reasoning capabilities, which will be of benefit to educators, curricular designers and researchers.