Enhancing TSH-based congenital hypothyroidism screening using machine learning and resampling algorithms
Bibliographic record
Abstract
PURPOSE: Congenital hypothyroidism (CH) is a common cause of severe intellectual disability, affecting approximately 1 in 2,000 newborns globally. Treatable with early intervention, congenital hypothyroidism has long been a target of newborn screening programs. Current thyroid stimulating hormone (TSH) based programs suffer from low positive predictive value, resulting in unnecessary diagnostic investigations. Congenital hypothyroidism screening has proven challenging for machine learning previously due to massive class imbalance and having a single well known predictor, preventing acceptable screening sensitivity. This study represents the most comprehensive evaluation of machine learning for congenital hypothyroidism screening to date. METHODS: Analyzing data from 616,910 infants screened by Newborn Screening Ontario between 2019 and 2024. 12 classification and 12 resampling algorithms were trained using 4 different optimization metrics, for a total of 576 distinct models evaluated using stratified 5-fold cross-validation to ensure robustness. Models were optimized for sensitivity and then positive predictive value using various metrics. Model explainability was assessed using SHAP values and feature importances. RESULTS: We were able to create a model achieving 16.8% PPV while maintaining 100% sensitivity using a RUSBoost classifier and Gaussian Noise resampling. This represents a 60% improvement in positive predictive value over the current approach. TSH remained the dominant predictor as in current screening, but our model was able to include minor amounts of additional information from other features to improve performance. CONCLUSION: These machine learning algorithms show no missed cases of CH and are able to significantly improve performance across robust testing. The findings suggest that machine learning offers a promising avenue for refining TSH-based CH screening processes, reducing false positives, and alleviating unnecessary stress and costs associated with current methods used by the majority of newborn screening programs globally.
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 | 0.001 |
| 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".