Machine learning and translational approaches to personalised care for women with gestational diabetes
LEAD SUPERVISOR: prof Jane Hirst, Nuffield Department of Women’s & Reproductive Health
Co-supervisor: Dr Nerys Astbury, Nuffield Department of Primary Care Health Sciences
Commercial partner: Sensyne Health plc, Oxford
Gestational diabetes mellitus (GDM) is a transient form of diabetes occurring during pregnancy. It is the commonest medical disorder of pregnancy, affecting 1 in 7 pregnant women worldwide and can lead to several problems, including excess fetal growth, problems during labour and birth, stillbirth, and neonatal complications such as hypoglycaemia. GDM is a heterogeneous condition, yet current care paradigms offer a “one size fits all” approach based on low-quality evidence. Individualising care for women could reduce unnecessary interventions, health anxiety, healthcare costs and improve outcomes and maternal satisfaction with care.
Sensyne Health (Sensyne) are a clinical artificial intelligence company based in Oxford. GDm-HealthTM was developed by the University of Oxford/OUH NHS Foundation Trust to improve remote monitoring and management of blood glucose in women with GDM. Sensyne licensed GDm-Health in 2018, and it is used by thousands of women per month in more than 50 NHS Trusts. Women submit blood glucose readings, tagged by mealtime with medication and diet comments through a secure mobile app, with a clinician facing, secure website for review and communication. Women give consent for the use of their anonymised data for research. Sensyne has created the world’s largest database of tagged blood glucose readings for women with GDM and is conducting groundbreaking research to improve care delivery. Sensyne and NDWRH are establishing a research hub of around ten hospitals using GDm-Health to facilitate sharing data on blood glucose metrics with outcomes at birth (due for completion in 2022). This national hub will build on our existing research collaborations through the RCOG Diabetes interest group.
During this DPhil, the student will be trained to consolidate a large dataset and use machine learning approaches for risk prediction in women with GDM. They will work with clinical experts and patient representatives to ensure clinical relevance and gain knowledge on the route to commercialisation for a new Software as a Medical Device (SaMD) in the UK.
1) Consolidate a dataset combining baseline maternal features and risk factors, glucose control and medication intake patterns during pregnancy, and perinatal outcomes using the GDm-Health database with NHS clinical data (with appropriate ethical approval).
2) Explore GDM-specific phenotypes based on outcomes at birth using feature selection methods such as Lasso.
3) Use machine learning approaches to develop and validate model/s to predict high-risk phenotypes at birth.
4) Work with the regulatory, legal and business development team at Sensyne to map the pathway towards regulatory approval and commercialisation of the product in the UK and other markets
5) Work with the clinical team to plan clinical validation and testing.
This project is well-aligned with the MRC remit, developing the student’s abilities in Prevention and Early detection/Stratified Medicine through the earlier detection and targeted treatment of women with high-risk GDM phenotypes, and Precision Medicine/New technologies and infrastructure supporting the use of big data generated through digital self-monitoring technologies to personalise care. The collaboration between Sensyne and NDWRH will enable translation into clinical practice.
Apply using course: DPhil in Women's and Reproductive Health