Data-Driven Modeling of Pregnancy-Related Complications

Espinosa, Camilo; Becker, Martin; Marić, Ivana; Wong, Ronald J.; Shaw, Gary M.; Gaudillière, Brice; Aghaeepour, Nima; Stevenson, David K.; Stelzer, Ina A.; Peterson, Laura S.; Chang, Alan L.; Xenochristou, Maria; Phongpreecha, Thanaphong; Francesco, Davide De; Katz, Michael; Blumenfeld, Yair J.; Angst, Martin S.

doi:10.1016/j.molmed.2021.01.007

Cited by 32 publications

(20 citation statements)

References 187 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Moreover, results showed that the use of models trained on data collected after the start of the pregnancy improved the performance by up to 5% on ROC-AUC, compared to models trained on the full patient EHR history, which includes both data before and after the start of the pregnancy. This supports previous findings in [41][42][43], which indicates that the most important risk factors are associated with events happening during the pregnancy timeline.…”

Section: Discussionsupporting

confidence: 92%

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: Discussionsupporting

confidence: 92%

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

“…Such integrative approaches might provide insight into how various stressors create strain, whether acute and resolvable, or chronic and intractable, with long-term health consequences. Advanced computational approaches, including knowledge graphs [20], neural networks [21], and ultimately, artificial general intelligence [22] might lead to the identification of stress-based modifiable factors that can steer biological systems (and, consequently, patient outcomes) without the need for costly pharmacologic or physiologic interventions. More precise interventions can improve patient outcomes and reduce high rates of maternal mortality, particularly among women of color [4,23].…”

Section: Accepted Manuscriptmentioning

confidence: 99%

Stress and Its Consequences—Biological Strain

et al. 2022

Self Cite

View full text Add to dashboard Cite

Understanding the role of stress in pregnancy and its consequences is important, particularly given documented associations between maternal stress and preterm birth and other pathologic outcomes. Physical and psychological stressors can elicit the same biological responses, known as biological strain. Chronic stressors, like poverty and racism (race-based discriminatory treatment), may create a legacy or trajectory of biological strain that no amount of coping can relieve in the absence of larger-scale socio-behavioral or societal changes. An integrative approach that takes into consideration simultaneously social and biological determinants of stress may provide the best insights into risk for preterm birth. The most successful computational approaches and the most predictive machine-learning models are likely to be those that combine information about the stressors and the biological strain (for example, as measured by different omics) experienced during pregnancy.

show abstract

“…10 However, EHR clinical data are complex, and difficult to interrogate. They are also heterogenous and lack standardization 11 . Recent computational advances help mitigate such limitations by data linkage and the availability of vast amounts of demographic, diagnostic, medication and clinical data 11,12 .…”

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…They are also heterogenous and lack standardization 11 . Recent computational advances help mitigate such limitations by data linkage and the availability of vast amounts of demographic, diagnostic, medication and clinical data 11,12 . Moreover, these data can often be retrieved at a fraction of the time and cost spent on prospective cohort studies or clinical trials and include thousands or tens of thousands of additional patients 12 . Here we have leveraged multi-task learning 16 and Long Short-Term Memory (LSTM) neural networks 15 to simultaneously predict the risk for the most important adverse neonatal outcomes throughout the pregnancy and after birth, using the entire maternal and newborn's medical history available in the EHR (Figure 1).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

AI-Driven Longitudinal Characterization of Neonatal Health and Morbidity

Francesco

Reiss

Roger

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

While prematurity is the single largest cause of death in children under 5 years of age, the current definition of prematurity, based on gestational age, lacks the precision needed for guiding care decisions. Here we propose a longitudinal risk assessment for adverse neonatal outcomes in newborns based on a multi-task deep learning model that uses electronic health records (EHRs) to predict a wide range of outcomes over a period starting shortly after the time of conception and ending months after birth. By linking the EHRs of the Lucile Packard Children’s Hospital and the Stanford Healthcare Adult Hospital, we developed a cohort of 22,104 mother-newborn dyads delivered between 2014 and 2018. This enabled a unique linkage between long-term maternal information and newborn outcomes. Maternal and newborn EHRs were extracted and used to train a multi-input multi-task deep learning model, featuring a long short-term memory neural network, to predict 24 different neonatal outcomes. An additional set of 10,250 mother-newborn dyads delivered at the same Stanford Hospitals from 2019 to September 2020 was used to independently validate the model, followed by a separate analysis of 12,256 mothers-newborn dyads at the University of California, San Francisco. Moreover, comprehensive association analysis identified multiple known and new associations between various maternal and neonatal features and specific neonatal outcomes. To date, this is the largest study utilizing linked EHRs from mother-newborn dyads and would serve as an important resource for the investigation and prediction of neonatal outcomes. An interactive website is available for independent investigators to leverage this unique dataset: https://maternal-child-health-associations.shinyapps.io/shiny_app/.

show abstract

Data-Driven Modeling of Pregnancy-Related Complications

Cited by 32 publications

References 187 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

Stress and Its Consequences—Biological Strain

AI-Driven Longitudinal Characterization of Neonatal Health and Morbidity

Contact Info

Product

Resources

About