Diagnostic statistical models are largely used in behavioral, educational, psychological, and health related applications to provide fine-grained information that can inform subsequent actions, such as instructions or interventions. These types of models are broadly defined, encompassing both latent class models and latent variable models, among others. Several decades of research have contributed to the development of various statistical models serving a variety of data types (e.g., binary, polytomous, continuous, etc.) and measurement scenarios (e.g., cross-sectional, longitudinal, large-scale, small-scale classroom assessment, etc.). In parallel, vast theoretical work has been conducted to establish model identification, and different algorithms have been developed to enable fast model estimation. Now, with the advancement of AI, such as large language models, it is expected that the diagnostic statistical models can be further expanded to accommodate high-dimensional data types (e.g., multimodal data, or process data) and be scaled up for real time feedback (e.g., adaptive learning platform).
In this special section, we invite authors to submit papers to advance integration of AI and diagnostic statistical models. Potential topics of interest include the following:
- Use of AI techniques (e.g., NLP, machine learning, Bayesian neural networks) to extract and validate Q-matrices or model latent attributes
- Hybrid models combining cognitive diagnostic models (CDMs) or multidimensional item response theory models (MIRTs) with deep learning, particularly for open-ended responses or process data
- Case studies showing how AI improves scalability, real-time diagnostics, or interpretability of CDMs or MIRTs in operational settings, such as in intelligent tutoring or adaptive learning systems
- Novel expansive applications of CDM in other domains, such as evaluating large language models (LLM) that are widely used in social and behavioral domains
This list is by no means exhaustive. We invite submissions on development and applications at the intersection of latent class models and AI. To be considered for this special section, manuscripts should have solid real data applications in psychological, educational, or social sciences. We especially encourage junior scholars to submit their research projects related to this topic.
Submission Guidelines
Interested authors are asked to submit a short proposal (1000 words or less) by October 31, 2025. The proposal should briefly describe the methodological contribution of the work on the integration of CDM and AI and the real data applications (in the relevant fields) of the proposed methodology. Note: That novel application of the extant, cutting edge methodology on interesting data sets is also encouraged. Submissions must represent original material that has neither been submitted to, nor published in, any other journal. The proposals will undergo initial review by the guest editors.
After reviewing the proposals, the guest editors will invite authors of highly rated abstracts to submit a full manuscript to the special section. This process is intended to ensure that submissions are aligned with the topics of the special section. The guest editors may also offer suggestions on the intended projects to ensure a good fit to the special section. The editors will notify all authors of their proposal decision by November 28, 2025.
Invited manuscripts must be submitted to the editorial manager submission system by clicking the button below:
Authors should select the special section “Leveraging AI to Empower Development and Application of Diagnostic Statistical Models” during the submission process. The deadline for submission of invited manuscripts is April 1, 2026.
All manuscripts submitted to the special section will go through the regular peer-review process (i.e., acceptance of the proposal does not guarantee publication). Please direct all queries regarding this special issue to the guest editors.
Guest Editors:
Dr. Chia-Yi Chiu
Measurement, Evaluation, & Statistics
Teachers College
Columbia University
Dr. Chun Wang
Measurement & Statistics
College of Education
University of Washington