Olga Demler, Ph.D.
Senior Scientist
- olga.demler@inf.ethz.ch
- Address
-
Biomedical Informatics Group
Universitätstrasse 6
8092 Zürich - Room
- CAB F52.2
I hold a joint appointment as a Senior Scientist Research at ETH Zürich in the Biomedical Informatics lab and as an Assistant Professor/Biostatistician at the Division of Preventive Medicine at MassGeneral Brigham Hospital / Harvard Medical School, Boston, USA.
At the BMI lab at ETH, I am expanding the methodological part of my research. I am interested in incorporating change-point detection when performing signal correction of metabolomics data. I also work on the assessment of the role of intransitive stochastic relationships in the analysis of medical data. Additionally, I work on expanding the capabilities of existing image data in disease diagnosis and prognosis using deep learning.
My research interests are three-fold: 1) conducting biomarker discovery studies using high throughput metabolomics and other -omics data; 2) building predictive models using Electronic Health Records and image data; 3) methodological research on building valid risk prediction models.
My clinical research focuses on the mechanistic pathways of cardiovascular disease development and progression, while my methodological research is applicable across medical fields.
I completed my Ph.D. at Boston University (Boston, USA), under the supervision of Ralph D’Agostino and Michael Pencina, working on methodological issues in the assessment of discrimination of risk prediction models. In 2012 I joined the Division of Preventive Medicine initially as a postdoc-level fellow and currently as an Assistant Professor, where I expanded my field of research by adding two more areas. I lead the design and analysis of high throughput metabolomics studies, primarily with Samia Mora’s group. I am also a Principal Investigator on a project adapting risk prediction models to clinical data. We have emulated a prospective cohort study by collecting medical history, laboratory biomarkers, and other data for 93K patients using Electronic Health Records data from large Boston-area hospitals with a median follow-up of 13 years. Recently we added image data to it.
Latest Publications
Abstract Question Can established cardiovascular risk tools be adapted for local populations without sacrificing interpretability? Findings This cohort study including 95 326 individuals applied a machine learning recalibration method that uses minimal variables to the American Heart Association’s Predicting Risk of Cardiovascular Disease Events (AHA-PREVENT) equations for a New England population. This approach strengthened the AHA-PREVENT risk equations, improving calibration while maintaining similar risk discrimination. Meaning The results indicate that the interpretable machine learning-based recalibration method used in this study can be implemented to tailor risk stratification in local health systems.
Authors Aniket N Zinzuwadia, Olga Mineeva, Chunying Li, Zareen Farukhi, Franco Giulianini, Brian Cade, Lin Chen, Elizabeth Karlson, Nina Paynter, Samia Mora, Olga Demler
Submitted JAMA cardiology
Abstract Fracture prediction is essential in managing patients with osteoporosis and is an integral component of many fracture prevention guidelines. We aimed to identify the most relevant clinical fracture risk factors in contemporary populations by training and validating short- and long-term fracture risk prediction models in 2 cohorts. We used traditional and machine learning survival models to predict risks of vertebral, hip, and any fractures on the basis of clinical risk factors, T-scores, and treatment history among participants in a nationwide Swiss Osteoporosis Registry (N = 5944 postmenopausal women, median follow-up of 4.1 yr between January 2015 and October 2022; a total of 1190 fractures during follow-up). The independent validation cohort comprised 5474 postmenopausal women from the UK Biobank with 290 incident fractures during follow-up. Uno’s C-index and the time-dependent area under the receiver operating characteristics curve were calculated to evaluate the performance of different machine learning models (Random survival forest and eXtreme Gradient Boosting). In the independent validation set, the C-index was 0.74 [0.58, 0.86] for vertebral fractures, 0.83 [0.7, 0.94] for hip fractures, and 0.63 [0.58, 0.69] for any fractures at year 2, and these values further increased for longer estimations of up to 7 yr. In comparison, the 10-yr fracture probability calculated with FRAX Switzerland was 0.60 [0.55, 0.64] for major osteoporotic fractures and 0.62 [0.49, 0.74] for hip fractures. The most important variables identified with Shapley additive explanations values were age, T-scores, and prior fractures, while number of falls was an important predictor of hip fractures. Performances of both traditional and machine learning models showed similar C-indices. We conclude that fracture risk can be improved by including the lumbar spine T-score, trabecular bone score, numbers of falls and recent fractures, and treatment information has a significant impact on fracture prediction.
Authors Oliver Lehmann, Olga Mineeva, Dinara Veshchezerova, HansJörg Häuselmann, Laura Guyer, Stephan Reichenbach, Thomas Lehmann, Olga Demler, Judith Everts-Graber, The Swiss Osteoporosis Registry Study Group
Submitted Journal of Bone and Mineral Research