Early Detection of Ovarian Cancer with Interpretable AI
Principal Investigator
Name
Weitong Huang
Degrees
Master of Philosophy
Institution
The Australian National University
Position Title
Student
Email
About this CDAS Project
Study
PLCO
(Learn more about this study)
Project ID
PLCO-933
Initial CDAS Request Approval
Mar 2, 2022
Title
Early Detection of Ovarian Cancer with Interpretable AI
Summary
Ovarian Cancer (OC), the fifth deadliest cancer among women, is often diagnosed at an advanced stage with poor prognosis due to being generally asymptomatic in early stages. It is believed with earlier detection, patient mortality rate can be significantly improved.
In this project, we would like to study serous bio-marker concentration patterns with computational science methods. Specifically, we would like to develop an interpretable machine learning algorithm for the early detection of OC, such that upon completion, our research result can be adopted by medical researchers in practice with high accessibility for screening and diagnosing.
Machine learning methods has shown great potential in data mining tasks. However, in research fields where explainability is as important as predictive behaviours, such as medical and public health, it often requires extra effort and knowledge to design and analyse methods that are being developed. We aim to develop machine learning algorithms with high specificity and sensitivity in this case for early detection of OC. Additionally, we will study, analyse, and incorporate interpretable AI techniques, those already existed and being developed, in a full data mining task life-cycle, to improve the explainability of our model so that they could be better accepted by medical practitioners with method accountability requirements.
In this project, we would like to study serous bio-marker concentration patterns with computational science methods. Specifically, we would like to develop an interpretable machine learning algorithm for the early detection of OC, such that upon completion, our research result can be adopted by medical researchers in practice with high accessibility for screening and diagnosing.
Machine learning methods has shown great potential in data mining tasks. However, in research fields where explainability is as important as predictive behaviours, such as medical and public health, it often requires extra effort and knowledge to design and analyse methods that are being developed. We aim to develop machine learning algorithms with high specificity and sensitivity in this case for early detection of OC. Additionally, we will study, analyse, and incorporate interpretable AI techniques, those already existed and being developed, in a full data mining task life-cycle, to improve the explainability of our model so that they could be better accepted by medical practitioners with method accountability requirements.
Aims
1. To improve predictive behaviour (measured by sensitivity and specificity) of ML algorithm for early detection of OC.
2. To study the interpretability and explainability techniques that can be used in medical research fields, specifically to benefit our study of OC.
3. To discover a systematic implementation of ML methods that are useful for clinical practices specifically in our study of OC.
Collaborators
Professor Amanda Barnard, The AUSTRALIAN NATIONAL UNIVERSITY
Associate Professor Hanna Suominen, The AUSTRALIAN NATIONAL UNIVERSITY
Mr Tommy Liu, The AUSTRALIAN NATIONAL UNIVERSITY
Mr Weitong Huang, The AUSTRALIAN NATIONAL UNIVERSITY
Related Publications
-
Explainable discovery of disease biomarkers: The case of ovarian cancer to illustrate the best practice in machine learning and Shapley analysis.
Huang W, Suominen H, Liu T, Rice G, Salomon C, Barnard AS
J Biomed Inform. 2023 Apr 14; Volume 141: Pages 104365 PUBMED