Ann-Kathrin Schalkamp

Applying Machine Learning for Healthcare

About Me

My research interest lies in the intersection of machine learning, medicine, and neuroscience. Given the advancements in technologies generating big data and methodological tools in the field of machine learning, medicine is transitioning into a data-driven era. This transition has the potential to revolutionise healthcare through improvements in early diagnosis, personalised treatment, and care. My goal is to contribute to this advancement through research in computational medicine/clinical informatics.

Curriculum Vitae (Updated September 2024)

Why clinical informatics?

Currently, diagnosis and treatment selection often rely on self-report and subjective decisions made by physicians. Augmenting their decision process by means of machine learning tools could largely fasten and objectify this process. An automatic, data-driven assessment of a patient’s health status can considerably alleviate the workload of physicians, enable early disease detection, and extract meaningful insights from the abundance of the medical data available.

Educational Background

Ph.D. (passed without corrections)
I completed a Ph.D. at Cardiff University and the UK Dementia Research Institute, under the guidance of Caleb Webber and Cynthia Sandor. My research addressed various gaps in Parkinson's disease research, ranging from risk prediction to disease monitoring, through the application of data-driven methods in deeply phenotyped cohorts. I worked with diverse data modalities, including digital sensors, electronic health records, genetics, brain imaging, blood biochemistry, and clinical data. My work involved the application of various machine learning techniques, both supervised and unsupervised, to extract insights and improve understanding in this field.

M.Sc. in Cognitive Science (4.0 GPA)
During my studies at the University of Tübingen, I prioritized expanding my methodological skills. To achieve this, I enrolled in advanced courses on machine learning, statistics, and data science. Notable highlights of my studies were seminars on computational psychiatry and deep learning for medicine, which demonstrated the significant impact advanced methodologies can have in the realm of neurological diseases.

B.Sc. in Cognitive Science (3.9 GPA)
Throughout my studies at the University of Osnabrück, I attended introductory courses covering various aspects of cognitive science, thereby gaining fundamental knowledge and skills in this interdisciplinary field. I developed a particular interest in neuropsychology, neurobiology, and neuroinformatics, and pursued advanced courses in these subjects. To establish a strong methodological foundation, I placed special emphasis on mathematics and computer science.

Research Experience

For a complete list of my publications, see my google scholar profile.

Postdoctoral Scholar at UCSF


I am using the electronic medical records from UC hospitals to improve clinical decision making. This involves extracting meaningful data from unstructured clinical notes, combining different data modalities, and training models to recommend treatments or predict future diagnoses.

Research Assistant/Associate at Imperial College London (now honorary position)


My projects revolve around the usage of digital sensors for neurodegenerative diseases, specifically risk prediction and monitoring of symptoms. Another project included the development of a large language model application for scientific publications.

PhD at Cardiff University


During my PhD I did data-driven research for Parkinson's disease including risk analysis, early diagnosis, progression modelling, and stratification with a focus on digital sensors data.

Master's Thesis at Neuronal Intelligence lab


During my time at the Neural Intelligence Lab, supervised by Fabian Sinz, I completed my master's thesis. The objective was to model the trajectories of cognitive and motor aging in healthy elderly individuals. By examining the influence of risk factors from various domains (genetics, lifestyle, tissue samples, demographics) on these trajectories, we aimed to establish general health guidelines and enable earlier diagnosis of pathological aging. I utilized advanced Bayesian reduced rank regression modeling to analyze longitudinal data from healthy elderly individuals with and without prodromal markers. Publication

Internship at Neuroscience of Motivation, Action, and Desire Laboratory


As part of my laboratory internship at NeuroMADLAB, I analysed the reliability of functional connectivity biomarkers for major depressive disorder using a multi-site dataset. This comprehensive analysis explored functional connectivity aberrations within and between large-scale brain networks in major depressive disorder through group-level analysis and individual-level analysis via supervised machine learning. By employing cross-site validation and assessing the similarity of group differences between sites, I shed light on the challenges of combining small sample size studies with different protocols.

Bachelor's Thesis


My bachelor's thesis investigated the suitability of electroencephalography (EEG) data acquired with the Traumschreiber, a portable high-tech sleep mask, for traditional EEG analysis and single-trial detection, as used in brain-computer interfaces (BCIs). This project involved designing, programming, conducting, and analyzing an EEG study. The main finding suggested the need to reconsider established EEG research standards, such as the placement of the reference electrode, for BCI applications. I proposed a novel approach of placing the reference electrode in the area where the signal is expected, which allowed for an increased sample size per participant without requiring additional trials. Short Summary of my Bachelor's Thesis