Development and validation of an explainable neural network model for predicting progression in type 2 diabetic kidney disease
-
By
-
Binfeng Xiong
-
Chengzheng Duan
-
Keying Lin
-
Sheng Xu
-
May 28, 2026
-
Clinical Scorecard: Creation and assessment of an interpretable neural network model for forecasting progression in type 2 diabetic kidney disease
At a Glance
| Category | Detail |
|---|
| Condition | Type 2 Diabetic Kidney Disease (T2DKD) |
| Key Mechanisms | Machine learning model using routine clinical parameters |
| Target Population | Patients diagnosed with T2DKD |
| Care Setting | Primary care settings |
Key Highlights
- T2DKD affects 20-40% of patients with type 2 diabetes mellitus.
- Four core predictors identified: MASLD, AST, DPN, and age.
- Neural network model achieved AUC of 0.742 and satisfactory calibration.
- Model provides a cost-effective tool for early intervention.
- Study conducted at Quzhou People’s Hospital, China.
Guideline-Based Recommendations
Diagnosis
- Diagnosis of T2DKD according to the 2025 APSN guideline.
Management
- Utilize machine learning models for predicting disease progression.
Monitoring & Follow-up
- Regular assessment of core predictors: MASLD, AST, DPN, and age.
Risks
- Risk of rapid progression to end-stage renal disease.
Patient & Prescribing Data
349 patients diagnosed with T2DKD.
Routine clinical parameters can effectively predict progression.
Clinical Best Practices
- Incorporate machine learning models in clinical practice for T2DKD.
- Focus on routinely available demographic data and laboratory parameters.
Related Resources & Content
