I'm passionate about research that uses technology to improve maternal health because women’s health tends to fall behind technological progress. Maternal health presents high socio-economic disparities, with most adverse outcomes in low-and-middle income countries, or among marginalized women in high-income countries, i.e., Black maternal health in the US, and indigenous maternal health in Australia.
A few years ago, while chatting with an OBGYN about AI's potential to help improve clinical workflow at a hospital in Ghana, he brought up the current challenges with Cardiotocography (CTG) interpretation and the need for limited highly-skilled CTG interpreters. I later found out that CTG interpretation, even in countries like the US, is still archaic, a similar pattern in several areas in women’s health.
We set out to research the challenging task of developing AI tools to interpret CTGs to serve as an early indicator of fetal wellbeing. As part of this work, we also sought to evaluate the robustness of the AI models to different factors like time when the signal was taken and patient clinical metadata, such as maternal age, and co-morbidites (pre-eclampsia, gestational diabetes, etc.). This blog post describes our methods and results and plans to open source our code as well for researchers to build on.
Many thanks to the wonderful team that made this happen, especially Nicole Chiou who worked passionately on this project while interning with me last year. Google Nicole Chiou Nichole Young-Lin, MD, MBA Katherine Heller Tiya Tiyasirichokchai Christopher Kelly Abdoulaye Diack Sanmi Koyejo Julie Cattiau
#womenshealth #maternalhealth #globalhealth #healthequity #artificialintellgence #SDGs #womenintech
Cardiotocography (CTG) is a technique used to monitor fetal well-being. Today we describe how ML models can assess CTGs to predict measures of fetal well-being to potentially assist healthcare providers, reducing burden & improving fetal outcomes. https://goo.gle/3BpfJnJ
Co-Founder & Chief Executive Officer at Beyond Lucid Technologies
6moThat's not correct. Want to write a rejoinder?