Qiuyu Pan, Nian Liu
AI education may overemphasize computational skills relative to frontline operational demands. Mitigation may require stratified pedagogy, real-world data, and less administration. Multi-stakeholder framework is valuable; causality requires further research.
OBJECTIVE: Evaluate supply-demand misalignment in public health artificial intelligence (AI) workforce education in Sichuan and Chongqing, inland Western China.
METHODS: Surveyed 1,031 stakeholders (150 employers, 680 students, 201 educators) using Importance-Performance Analysis (IPA) and Latent Profile Analysis (LPA) to quantify skill deficits. Multivariate models assessed collaborative training and faculty transfer factors, guided by a conceptual framework integrating demand, supply, and training perspectives. All coefficients are associational, not causal.
RESULTS: IPA showed employers prioritised risk assessment; students focused on algorithmic construction. LPA on six practical skills identified two profiles: High-Order Application Group (23.2%) and Foundation-Weak Group (76.8%); the weighted combination of profile means reconciled with the overall sample mean. Employers' deficit perception was positively associated with their collaboration willingness (β = 0.235, 95% CI [0.061, 0.410], p < 0.01). Institutional innovation negatively moderated the link between faculty AI proficiency and research mentorship (β = -0.120, [-0.216, -0.024], p < 0.05).
CONCLUSION: AI education may overemphasize computational skills relative to frontline operational demands. Mitigation may require stratified pedagogy, real-world data, and less administration. Multi-stakeholder framework is valuable; causality requires further research.