Validation of Artificial Intelligence (AI)-Based Kidney Disease Progression Prediction (KDPP) Models in the US Population
Bibliographic record
Abstract
Background: AI and big data are revolutionizing personalized CKD management. KDPP models use deep learning to risk stratify patients for optimal treatment planning. Initially validated in Taiwan and granted FDAbreakthrough device designation, this study tests KDPP's generalizability in U.S. using NIDDK's CRIC database. Methods: KDPP includes 2 deep-learning models built using 9,529 CKD stage 3-5 patients from CMUH Taiwan: KDPP-RP for predicting rapid disease progression and KDPP-IR for forecasting renal replacement therapy (RRT). KDPP-IR was validated with CRIC (4,465 participants); KDPP-RP validation was limited by data scarcity. The models categorize patients into risk tiers (low, moderate, high) and are evaluated using AUC, sensitivity, specificity, and PPV/NPV. Results: The CMUH cohort is older, has lower median eGFR (27 vs. 41.9 mL/min/1.73m2) than CRIC. KDPP-IR accurately predicted RRT with AUCs of 0.96 for CMUH and 0.89-0.92 for CRIC, performing well across races (Table 1). KDPP-IR's sensitivity and precision slightly decreased, but specificity improved. It reliably identified low-risk CRIC patients with NPVs of 0.99-1.00, though PPVs for high-risk varied (0.31-0.79). Kaplan-Meier curves showed distinct risk stratification, and calibration plots showed better agreement between observed and predicted risks than KFRE. (Figure 1). Conclusion: The validation demonstrated robust AUCs and generalizability for risk prediction. Future enhancements will optimize both models for diverse ethnic groups to broaden their applicability. Funding: Commercial Support - AWAK Technologies
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".