Dense phenotyping from electronic health records enables machine-learning-based prediction of preterm birth

Abraham, Abin; Le, Brian; Kosti, Idit; Straub, Péter; Edwards, Digna R. Velez; Davis, Lea K.; Muglia, Lou; Rokas, Antonis; Bejan, Cosmin Adrian; Sirota, Marina; Capra, John A.

doi:10.1101/2020.07.15.20154864

Cited by 8 publications

(4 citation statements)

References 87 publications

(152 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A recent work [25], which predicts preterm birth using gradient boosted decision trees on EHR data, have achieved an a ROC-AUC of 0.75 and PR-AUC of 0.40 at 28 weeks of gestation (which is approximately two months before delivery). Compared to PredictPTB, this model provides predictions starting from two months before delivery and up to 10 days before delivery, and no information was provided about the model's performance at earlier stages of the pregnancy timeline.…”

Section: Discussionmentioning

confidence: 99%

“…As a result, machine learning is of great interest to better predict PTB and several studies have attempted to predict PTB using machine learning techniques on a set of pre-defined clinical risk factors [13][14][15][16][17][18][19][20], or leveraging diverse variables from electronic health record (EHR) data [21]. More recently, few studies, with promising results, have used deep learning techniques to predict PTB using ultrasound and MRI images [22,23], and high-dimensional EHR data [24,25].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

PredictPTB: An Interpretable Preterm Birth Prediction Model Using Attention-based Recurrent Neural Networks

AlSaad

Malluhi

Boughorbel

2021

Preprint

View full text Add to dashboard Cite

Background: Early identification of pregnant women at risk for preterm birth (PTB), a major cause of infant mortality and morbidity, has a significant potential to improve prenatal care. However, we lack effective predictive models which can accurately forecast PTB and complement these predictions with appropriate interpretations for clinicians. In this work, we introduce a clinical prediction model (PredictPTB) which combines variables (medical codes) readily accessible through electronic health record (EHR) to accurately predict the risk of preterm birth at 1, 3, 6, and 9 months prior to delivery. Methods: The architecture of PredictPTB employs recurrent neural networks (RNNs) to model the longitudinal patient's EHR visits and exploits a single code-level attention mechanism to improve the predictive performance, while providing temporal code-level and visit-level explanations for the prediction results. We compare the performance of different combinations of prediction time-points, data modalities, and data windows. We also present a case-study of our model's interpretability illustrating how clinicians can gain some transparency into the predictions. Results: Leveraging a large cohort of 222,436 deliveries, comprising a total of 27,100 unique clinical concepts, our model was able to predict preterm birth with an ROC-AUC of 0.82, 0.79, 0.78, and PR-AUC of 0.40, 0.31, 0.24, at 1, 3, and 6 months prior to delivery, respectively. Results also confirm that observational data modalities (such as diagnoses) are more predictive for preterm birth than interventional data modalities (e.g., medications and procedures). Conclusions: Our results demonstrate that PredictPTB can be utilized to achieve accurate and scalable predictions for preterm birth, complemented by explanations that directly highlight evidence in the patient's EHR timeline.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

PredictPTB: An Interpretable Preterm Birth Prediction Model Using Attention-based Recurrent Neural Networks

AlSaad

Malluhi

Boughorbel

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…ML-based model is also being used for predicting a lot of factors of childbirth. For example, Abraham, Abin, et al described a new technique for gathering various information from EHRs in order to predict singleton preterm birth by applying various machine learning models [7]. Recently, Islam, Muhammad Nazrul, et al has presented research regarding childbirth mode with two-fold findings: first, the potential highlights for deciding the method of labor, and second, machine learning algorithms for anticipating the suitable way of labor (vaginal birth, crisis cesarean, cesarean birth) [8].…”

Section: Related Workmentioning

confidence: 99%

Predicting the Appropriate Mode of Childbirth using Machine Learning Algorithm

Kowsher¹,

Prottasha²,

Tahabilder³

et al. 2021

IJACSA

View full text Add to dashboard Cite

A woman's satisfaction with childbirth may have immediate and long-term effects on her health as well as on the relationship with her newborn child. The mode of baby delivery is genuinely vital to a delivery patient and her infant child. It might be a crucial factor for ensuring the safety of both the mother and the child. During the baby delivery, decision-making within a short time becomes very challenging for the physician. Besides, humans may make wrong decisions selecting the appropriate delivery mode of childbirth. A wrong decision increases the mother's life risk and can also be harmful to the newborn baby's health. Computer-aided decision-making can be an excellent solution to this problem. Considering this scope, we have built a supervised machine learning-based decision-making model to predict the most suitable childbirth mode that will reduce this risk. This work has applied 32 supervised classifier algorithms and 11 training methods on the real childbirth dataset from the Tarail Upazilla Health complex, Kishorganj, Bangladesh. We have also analyzed the result and compared them using various statistical parameters to determine the bestperformed model. The quadratic discriminant analysis has shown the highest accuracy of 0.979992 with the F1 score of 0.979962. Using this model to decide the appropriate labor mode may significantly reduce maternal and infant health risks.

show abstract

“…This type of dataset provides a great opportunity to deeply investigate diseases and identify associations to facilitate understanding disease prevention and progression. Recently, EMR has been utilized for other diseases for creating comorbidity networks 27 , identifying disease subtypes 28 and predicting disease outcomes 29, 30 highlighting the potential of utilizing EMR data to extract insight and utility for complex and heterogeneous diseases 31 , but a big data integrative analysis with EMR data has not yet been applied to the AD phenotype.…”

Section: Introductionmentioning

confidence: 99%

Deep clinical phenotyping of Alzheimer’s Disease Patients Leveraging Electronic Medical Records Data Identifies Sex-Specific Clinical Associations

Tang

Oskotsky

Mantyh

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Alzheimer's Disease (AD) is a devastating disorder that is still not fully understood. Sex modifies AD vulnerability, but the reasons for this are largely unknown. There has been efforts to understand select comorbidities, covariates, and biomarkers of AD, with and without sex stratification - but there has not yet been an integrative, big data approach to identify clinical and sex specific associations with AD in an unbiased manner. Electronic Medical Records (EMR) contain extensive information on patients, including diagnoses, medications, and lab test results, providing a unique opportunity to apply phenotyping approaches to derive insights into AD clinical associations. Here, we utilize EMRs to perform deep clinical phenotyping and network analysis of AD patients to provide insight into its clinical characteristics and sex-specific clinical associations. We performed embeddings and network representation of patient diagnoses to visualize patient heterogeneity and comorbidity interactions and observe greater connectivity of diagnosis among AD patients compared to controls. We performed enrichment analysis between cases and controls and identified multiple known and new diagnostic and medication associations, such as positive associations with AD and hypertension, hyperlipidemia, anemia, and urinary tract infection - and negative associations with neoplasms and opioids. Furthermore, we performed sex-specific enrichment analyses to identify novel sex-specific associations with AD, such as osteoporosis, depression, cardiovascular risk factors, and musculoskeletal disorders diagnosed in female AD patients and neurological, behavioral, and sensory disorders enriched in male AD patients. We also analyzed lab test results, resulting in clusters of patient phenotype groups, and we observed greater calcium and lower alanine aminotransferase (ALT) in AD, as well as abnormal hemostasis labs in female AD. With this method of phenotyping, we can represent AD complexity, and identify clinical factors that can be followed-up for further temporal and predictive analysis or integrate with molecular data to aid in diagnosis and generate hypotheses about disease mechanisms. Furthermore, the negative associations can help identify factors that may decrease likelihood of AD and help motivate future drug repurposing or therapeutic approaches.

show abstract

Dense phenotyping from electronic health records enables machine-learning-based prediction of preterm birth

Cited by 8 publications

References 87 publications

PredictPTB: An Interpretable Preterm Birth Prediction Model Using Attention-based Recurrent Neural Networks

PredictPTB: An Interpretable Preterm Birth Prediction Model Using Attention-based Recurrent Neural Networks

Predicting the Appropriate Mode of Childbirth using Machine Learning Algorithm

Deep clinical phenotyping of Alzheimer’s Disease Patients Leveraging Electronic Medical Records Data Identifies Sex-Specific Clinical Associations

Contact Info

Product

Resources

About