Machine Learning Risk Prediction for Treated Retinopathy of Prematurity in Infants
Henry P Foote1, Yanchen Jessie Ou2, Suchir Bhatt3
1Department of Pediatrics, Duke University, Durham, North Carolina, USA.
Insights
Machine learning models can identify infants needing retinopathy of prematurity (ROP) treatment, potentially reducing unnecessary screenings. These models offer a more precise approach to ROP detection in high-risk infants.
Area of Science:
- Neonatal ophthalmology
- Medical artificial intelligence
- Predictive analytics in healthcare
Background:
- Retinopathy of prematurity (ROP) is a primary cause of childhood blindness.
- Current ROP screening guidelines may be too broad, leading to unnecessary evaluations.
- There is a need for improved models to identify infants at high risk for ROP.
Purpose of the Study:
- To develop and validate machine learning (ML) models for predicting the need for ROP treatment.
- To stratify infants based on ROP treatment timing using ML models.
- To compare the performance of ML models against traditional logistic regression (LR) models.
Main Methods:
- Utilized a multicenter cohort of 103,701 infants (birth weight ≤1,500g or gestational age ≤30 weeks).
- Developed ML models at 2-week intervals from postnatal day 14 to 98 using clinically relevant variables.
- Validated models in a separate cohort of 25,105 infants and compared performance to an LR model.
Main Results:
- The day 28 ML model demonstrated superior performance over the LR model in the validation cohort (AUROC: 0.916 vs. 0.903; AP: 0.190 vs. 0.160).
- At a 100% sensitivity threshold, the ML model achieved a negative predictive value of >99.9%.
- The ML model could potentially reduce the number of infants requiring screening by 14% compared to current guidelines.
Conclusions:
- ML models are effective in predicting the need for ROP treatment and stratifying infant risk.
- These models show potential for reducing unnecessary ROP screenings.
- Further research is required to implement these model-based ROP predictions in clinical practice.
Introduction:
Retinopathy of prematurity (ROP) is a leading cause of childhood blindness. However, current screening guidelines may be overly broad, necessitating better models to detect high-risk infants.
Methods:
From a multicenter cohort of 103,701 infants (3,301 [3.2%] treated for ROP) discharged from 298 neonatal intensive care units from 2006 to 2017 with birth weight ≤1,500 grams or gestational age ≤30 weeks, we used clinically relevant variables to develop machine learning (ML) models at 2-week intervals from postnatal day 14 to 98 to stratify infants by ROP treatment timing. We assessed model performance by concordance index, area under the receiver operating characteristic curve (AUROC), and average precision (AP), validated performance in a cohort of 25,105 infants across 231 sites from 2018 to 2020, and compared model performance to a logistic regression (LR) model.
Results:
In the validation cohort, the day 28 ML model outperformed the LR model by AUROC (0.916 [0.905-0.926] vs. 0.903 [0.892-0.914]; p < 0.001) and AP (0.190 [0.167-0.217] vs. 0.160 [0.140-0.183]; p < 0.001). Using the ML model at a 100% sensitivity threshold would have negative predictive value of >99.9% and could reduce the number of infants needing screening by 14% compared to current guidelines.
Conclusion:
ML models can effectively predict the need for ROP treatment and stratify infants by risk, potentially reducing unneeded screening. Future work is needed to translate model-based ROP predictions to the clinical setting.
