AI-Powered Problem- and Case-based Learning in Medical and Dental Education: A Systematic Review and Meta-analysis

Abstract

Introduction and aims: Advances in artificial intelligence (AI) technology have generated a revolution in medical and dental education, which may offer promising solutions to tackle the challenges of traditional problem-based learning (PBL) and case-based learning (CBL). The objective of this study was to assess the available evidence concerning AI-powered PBL/CBL on students' knowledge acquisition, clinical reasoning capability and satisfaction.

Methods: An electronic search was carried out on PubMed, MEDLINE, the Cochrane Central Register of Controlled Trials and Web of Science. Clinical trials published in English with full text available, which implemented AI technologies in PBL/CBL in the medical/dental field and evaluated knowledge acquisition, clinical reasoning and/or satisfaction were included. The quality assessment was conducted using RoB 2 by two calibrated assessors. Data synthesis and meta-analysis were performed, the standardised mean difference (SMD) or standardised mean (SM) and 95% confidence intervals (CIs) were calculated, and heterogeneity was quantified.

Results: Six randomized controlled trials were included, with an overall risk of bias judged to have 'some concerns'. For knowledge acquisition, 4 studies were included in the meta-analysis. A low heterogeneity (I² = 20%) was detected and a fixed-effect model was utilised. Compared with the control group, the AI intervention significantly improved knowledge acquisition by 46% (95% Cls [0.18-0.73], P = .001). For clinical reasoning capability, due to methodological and measurement heterogeneity among studies, statistical analysis was not feasible. Three studies were selected for the meta-analysis of students' satisfaction. Heterogeneity was moderate (I² = 32%), and a generic inverse variance method was selected. The pooled SM score was 0.7 (95% Cls [0.47-0.92]), and the overall effect was statistically significant (P < .00001).

Conclusion: Despite limitations such as the limited number of included studies and the overall risk of bias concerns, AI-powered PBL/CBL has the potential to enhance students' knowledge acquisition and learner satisfaction compared to traditional learning approaches.

Clinical relevance: Not applicable.

Keywords: Artificial intelligence; Case-based learning; Intelligent tutoring system; Meta-analysis; Problem-based learning.

Fig. 1

A PRISMA flowchart illustrating the article screening process.

Fig. 2

The risk of bias in each study included in this systematic review. D1: bias arising from the randomization process; D2: bias due to deviations from intended interventions; D3: bias due to missing outcome data; D4: bias in measurement of the outcome; D5: bias in selection of the reported result. The green, yellow, and red colours represent a low risk of bias, some concerns, and a high risk of bias, respectively. CBL, case-based learning; COMET, COllaborative MEdical Tutor; IIAS, intelligent information access system; ITS, intelligent tutoring system; PBL, problem-based learning; WFO, Watson for Oncology.

Fig. 3

A forest plot of meta-analysis of the studies comparing students’ knowledge acquisition in AI-powered PBL/CBL with PBL/CBL without AI integration or traditional lectures. The box represents the point estimate of the SMD of the study, the horizontal line indicates the 95% CI, and the size of the box reflects the weight of the study in relation to the pooled estimate. The diamond represents the overall effect estimate derived from the meta-analysis.

Fig. 4

A forest plot of the studies comparing clinical reasoning capability of AI-powered PBL/CBL with PBL/CBL without AI integration or traditional lectures. The box represents the point estimate of the SMD of the study, the horizontal line indicates the 95% CI, and the size of the box reflects the weight of the study in relation to the pooled estimate.

Fig. 5

A forest plot of a meta-analysis of the studies comparing students’ satisfaction with AI-powered PBL/CBL. The box represents the point estimate of the SM of the study, the horizontal line indicates the 95% CI, and the size of the box reflects the weight of the study in relation to the pooled estimate. The diamond represents the overall effect estimate derived from the meta-analysis.