What the study found
The study found that a deep learning framework combining family background and psychological evaluation indicators classified student GPA with the best performance when using TabTransformer plus a feature-gating mechanism. It also found negative associations between GPA and some psychological measures, including depression and anxiety.
Why the authors say this matters
The authors conclude that the framework may help identify academic risk and inform targeted academic assistance and psychological interventions. They suggest that using family background and psychological factors together can support this kind of classification in the post-pandemic era.
What the researchers tested
The researchers collected data from 1,692 undergraduates at a Chinese university. The dataset included family background factors, SCL-90 psychological evaluation scores, and GPA records, and they compared four deep learning models: TabTransformer, DCNv2, AutoInt, and MLP-ResNet, with a lightweight feature-gating mechanism added to improve feature selection.
What worked and what didn't
TabTransformer with the gating mechanism performed best, with an Accuracy of 0.798 and an AUC of 0.833. GPA was significantly negatively correlated with SCL-90 domains including depression and anxiety, and less favorable family background factors such as lower economic status and longer left-behind years were correlated with poorer psychological assessment outcomes.
What to keep in mind
The abstract describes data from one Chinese university, so the scope is limited to that sample. No other limitations are described in the available summary.
Key points
- The best-performing model was TabTransformer with a gating mechanism.
- Its reported performance was Accuracy 0.798 and AUC 0.833.
- GPA was negatively correlated with SCL-90 domains including depression and anxiety.
- Lower family economic status and longer left-behind years were linked to poorer psychological assessment outcomes.
- The study used data from 1,692 undergraduates at a Chinese university.
Disclosure
- Research title:
- Deep learning classified GPA using family and psychological factors
- Authors:
- Hongrong Zhang, Fang Fang, Yi Wang, Yong Huang, Ya Li
- Institutions:
- Anhui University of Traditional Chinese Medicine, Anhui University of Traditional Chinese Medicine, Anhui University of Traditional Chinese Medicine, Anhui University of Traditional Chinese Medicine, Anhui University of Traditional Chinese Medicine
- Publication date:
- 2026-03-05
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with scholarly research without reading academic papers — one filtered digest, every Friday.