Birth Hypoxia, Birth Outcome, Adverse, Birth, Preterm, Birth Weight, Diabetes Complications, Gestational Complication, Gestational Diabetes, Gestational Diabetes Mellitus in Pregnancy, Gestational Hypertension, Gestational Mother, Gestational Weight Gain, Pregnancy Bleeding, Pregnancy Complications, Pregnancy, High Risk, Pregnancy in Diabetic, Pregnancy Induced Hypertension, Pregnancy Loss, Pregnancy Preterm, Pregnancy Weight Gain
Conditions
Keywords
gestational diabetes, machine learning, predictive monitoring, clustering, patient subtype, deep learning, transfomer model, foundation model, diabetes, blood glucose, c section, high risk pregnancy, diabetes complications
Brief summary
The primary goal is to predict the clinical outcomes of mother and baby using blood glucose and other routinely collected clinical data in pregnancy to predict adverse outcomes at birth in women with GDM. The secondary goal is to develop models to predict optimal blood glucose testing schedules for pregnant women. Exploratory Objectives are (1) to understand patterns of dosage and / or medication choice and (2) to describe different phenotypes of gestational diabetes based on multiple data input.
Detailed description
Gestational diabetes is a sub-type of diabetes that causes a person's blood sugar level to become too high during pregnancy. This health condition affects approximately 10% of pregnant women in the UK and up to 20% worldwide. Women who have gestational diabetes need to take daily blood tests to monitor their blood sugar. While much work exists on telehealth using blood glucose monitoring, little exists in modern AI-based methods for performing the prediction of patient health status in such settings. This study builds on world-leading research in this field within the Institute of Biomedical Engineering and the Nuffield Department of Women's & Reproductive Health at the University of Oxford. The focus of this project is to clearly identify patients in different risk groups, predict the clinical outcome of mothers and babies, and reduce the overall number of blood tests. During this study, CI and investigators will develop novel state-of-the-art AI models to improve blood glucose control. This study will use existing retrospective data in pursuit of objectives. The hypothesis in this study is that better blood glucose control will improve clinical outcomes. The predictive models developed in this research study will provide an estimate of patient-specific health risk through time, and notify patients of the clinically appropriate number of blood glucose tests required to monitor their condition. As a result, innovations arising from this study can support future studies to facilitate rapid clinical treatment, transform a hospital-only treatment pathway into a cost-effective home-based alternative, and improve the overall quality of maternal healthcare.
Interventions
None listed
Sponsors
Royal Academy of Engineering
Oxford University Hospitals NHS Trust
University of Oxford
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
FEMALE
Age
18 Years to 99 Years
Healthy volunteers
No
Inclusion criteria
* Pregnant women with GDM during pregnancy * Record of blood glucose monitoring registered on the GDm-Health system
Exclusion criteria
The participant may not enter the study if ANY of the following apply: * Women who have not consented for their data to be shared through GDm-Health * Women who opted out of the use of their data in health research
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Clinical outcome of mothers at birth | From enrollment to the delivery date, assesed upto 52 weeks. | Gestational age at delivery in years, Mode of delivery (vaginal, caesarean, assisted) as categocial measurements, Maternal weight gain in kg, Maternal pregnancy-induced hypertension in binary Yes or No, Maternal pregnancy-induced pre-eclampsia in binary Yes or No. |
| Clinical outcome of mothers after birth | From the delivery date to the date that mother was discharged from the hospital, assessed up to 52 weeks, whichever come first. | Perineal trauma (3rd- or 4th-degree perineal tear or tear requiring suturing in the operating room) in text notes, Admission to higher level of care for mother in ICD codes, Length of hospital stay for mother in days, Method of feeding at discharge from hospital in texts. |
| Clinical outcome of neonates at birth | From enrollment to the delivery date to the date to giving birth, assesed upto 52 weeks. | Newborn status at birth in categocial (alive, stillbirth), Birth weight in kg, Gender of neonate in categocial of male or female, APGAR score at 5 mins in numerical numbers from 0 to 10. |
| Clinical outcome of neonates after birth | From the delivery date to the date that neonate was discharged from the hospital, assessed up to 52 weeks, whichever come first. | Incidence of shoulder dystocia/birth injury in ICD codes to indicate yes or no, Incidence of neonatal hypoglycaemia in ICD codes to indicate Yes or No, Incidence of neonatal significant hyperbilirubinemia in ICD codes to indicate Yes or No, Length of hospital stay for neonate in days. |
Countries
United Kingdom
Outcome results
None listed