Research Project: Advancing Statistical and Mathematical Validation for Maternal Mental Health Recovery Pathway: Deep Learning, Evaluation, and Evidence-Based Insights
Why did you apply for this internship?
The internship appealed to me as it offered a chance to apply theoretical data science techniques in a real-world context that has potential for meaningful community impact.
What did you hope to gain in completing this project?
I hoped to gain an appreciation for how things are done in a professional setting. Getting insights from veterans of the industry, as well as understanding the considerations made when dealing with sensitive data.
Project Overview
This internship project develops innovative deep learning methodologies with statistical validation to improve early detection of postnatal depression, addressing gaps in traditional screening methods.
Using a synthetic dataset of 10,000 patients across 100,000 clinical visits, the project creates predictive models for earlier identification of at-risk mothers. A complementary web application will visualise patient journey outlooks, providing healthcare providers with actionable insights into risk profiles and care trajectories.
This research aims to enhance maternal mental health outcomes in the North East and North Cumbria region by combining cutting-edge machine learning with user-centred design to transform postnatal depression screening and intervention approaches.
What were the key results of your research project?
The study found that geographic factors like coastal proximity had no significant link to mental health scores, but socioeconomic deprivation was concentrated in specific urban and rural areas.
The ANNs demonstrated strong predictive performance with an accuracy of 0.866–0.867, outperforming other models, and SHAP analysis revealed that the most influential predictors of postnatal depression were inflammatory index, BMI, and PHQ-9 scores.
This work shows that automated risk assessment tools are feasible for perinatal care, and that individualized, patient-centred factorsare more predictive of maternal mental health outcomes than geographic location.
How do you feel you have benefited from completing this internship and has it made you consider future career paths?
The internship has given me an invaluable look into how labs work and how they expect the people in them to work. I got to see the timelines for projects, sit in on meetings with potential clients and got countless valuable insights from my supervisor about industry practices.
It's made me consider being a researcher, whether that be in a university lab as a doctoral candidate or in industry.
Download presentation slides: