Rapid clinical antimicrobial resistance prediction
In my PhD project, I developed models predicting antimicrobial resistance from MALDI-TOF mass spectrometry data, through topological data analysis, kernel methods and optimal transport. These models have the potential to give clinical decision support in the early stages of an bacterial or fungal infection, thereby improving patient care by the reducing time to an informed treatment decision.
MALDI-TOF mass spectrometry has been adopted in routine clinical microbiology to identify bacterial and funghi specimen. This technology is inexpensive and delivers MALDI-TOF mass spectra rapidly after sample collection. The spectra also harbour the potential to contain information beyond microbial indentity. My research focuses on using Machine Learning method to extract this additional information and make it accessible to patient care.
Recent work of mine on this topic includes a Nature Medicine article on Direct antimicrobial resistance prediction from clinical MALDI-TOF mass spectra using machine learning, an article on Topological and kernel-based microbial phenotype prediction from MALDI-TOF mass spectra accepted at ISMB 2020, published at OUP Bioinformatics and Machine learning for microbial identification and antimicrobial susceptibility testing on MALDI-TOF mass spectra: a systematic review published in Clinical Microbiology and Infection.
With an interdisciplinary background in microbiology, structural physics and bioinformatics, I'm interested in solving real-world problems at the interphase of Machine Learning and Healthcare. Beyond my core PhD project, I've worked on applying survival analysis to assess disease risk from human genotype data, and topological data analysis for single-cell cancer data.