Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Master Thesis on Disease Modeling and Analysis in SCI Comorbitities
The goal of this work is to create algorithms for exploring multiple existing and new models of the body conditions from data of spinal cord injured patients with a collaborative project between the SPZ-ETHZ.
You would join a team of clinical and research scientists in the task of improving the long-term prognosis of Spinal Cord Injury (SCI) through physiological and clinical data.
We will make use of the data at the SPZ for exploring available dimensions through analysis of correlations, causal models, regression and classification models of common functional disorders and diseases for SCI patients, such as pressure injuries, and autonomic dysreflexia.
Keywords: Machine learning, Data science, Medical and health science, computing and computational science, engineering and technology, information, time-series, real-world data, prediction
ETHZ-SPZ Master Thesis at the Sensory-Motor Systems lab of ETH-Zurich in collaboration with the Swiss Paraplegic Center (SPZ) at Nottwil.
Pressure Injuries are sores or ulcers that happen on areas of the skin that are under pressure. Among the SCI (Spinal Cord Injuries) population, this is a highly prevalent co-morbidity because of the long time spent in a wheelchair or a bed without pressure release. Due to the loss of sensation, persons with SCI do not perceive the corresponding warning signs such as pain thus their PI could escalate to many severe stages without any intervention. Despite much research on PI, they are a complex process still unclear. The combination of shear and friction, the time and frequency of applied pressure, and the soft tissue’s characteristics and oxygenation level are significant factors. Nonetheless, an individual’s medical history, etiology of SCI, muscle build, posture, nutritional status, and many other behavioural factors are also significant contributing factors that must be considered. Our objective is to identify the causal relations among those risk factors towards the onsets and development of PI in different locations, through clinical data. We would test both constraint-based methods and score-based methods for causal discovery and develop an interactive system for doctors to input their expert knowledge and edit the learned graph from data.
Causal learning has attracted growing attention in both academia and industry since finding the underlying causal relations and making causal inferences are fundamental to various disciplines of science [1]. As a rigorous statistical modelling framework, graphical models can represent the relations of variables in a compact and organized manner, which become especially popular in fields of healthcare and biomedical research where an explicit representation of knowledge is expected.
[1] Glymour, C., Zhang, K., & Spirtes, P. (2019). Review of causal discovery methods based on graphical models. Frontiers in genetics, 10, 524.
ETHZ-SPZ Master Thesis at the Sensory-Motor Systems lab of ETH-Zurich in collaboration with the Swiss Paraplegic Center (SPZ) at Nottwil.
Pressure Injuries are sores or ulcers that happen on areas of the skin that are under pressure. Among the SCI (Spinal Cord Injuries) population, this is a highly prevalent co-morbidity because of the long time spent in a wheelchair or a bed without pressure release. Due to the loss of sensation, persons with SCI do not perceive the corresponding warning signs such as pain thus their PI could escalate to many severe stages without any intervention. Despite much research on PI, they are a complex process still unclear. The combination of shear and friction, the time and frequency of applied pressure, and the soft tissue’s characteristics and oxygenation level are significant factors. Nonetheless, an individual’s medical history, etiology of SCI, muscle build, posture, nutritional status, and many other behavioural factors are also significant contributing factors that must be considered. Our objective is to identify the causal relations among those risk factors towards the onsets and development of PI in different locations, through clinical data. We would test both constraint-based methods and score-based methods for causal discovery and develop an interactive system for doctors to input their expert knowledge and edit the learned graph from data.
Causal learning has attracted growing attention in both academia and industry since finding the underlying causal relations and making causal inferences are fundamental to various disciplines of science [1]. As a rigorous statistical modelling framework, graphical models can represent the relations of variables in a compact and organized manner, which become especially popular in fields of healthcare and biomedical research where an explicit representation of knowledge is expected.
[1] Glymour, C., Zhang, K., & Spirtes, P. (2019). Review of causal discovery methods based on graphical models. Frontiers in genetics, 10, 524.
1. Preliminary Research on Causal Discovery: You will review and study classic and state-of-the-art causal discovery approaches for clinical data (with mixed-type variables) and methods that incorporate expert knowledge.
2. Data Exploration and Analysis: You will process and analyze the private datasets gathered from different patient groups divided according to their purposes of hospital stay, which have the same feature space consisting of mainly demographic features and high-level features extracted from bio-signals. We will also explore some public datasets with mixed-type variables from other fields such as heart diseases and earth science.
3. Methodology Development and Deployment: You will help develop and implement an iterative algorithm for causal discovery with a human in the loop against benchmarks on the simulated and real datasets. Based on the methodology, an interactive web interface for doctors to input their expert knowledge and improve the learning results. If possible, you will assist in deriving the mathematical guarantee for the algorithm in convergence and stability.
4. Visualization and Presentation: You will prepare a manuscript for publication with a clear and engaging representation of the empirical results (optional).
1. Preliminary Research on Causal Discovery: You will review and study classic and state-of-the-art causal discovery approaches for clinical data (with mixed-type variables) and methods that incorporate expert knowledge. 2. Data Exploration and Analysis: You will process and analyze the private datasets gathered from different patient groups divided according to their purposes of hospital stay, which have the same feature space consisting of mainly demographic features and high-level features extracted from bio-signals. We will also explore some public datasets with mixed-type variables from other fields such as heart diseases and earth science. 3. Methodology Development and Deployment: You will help develop and implement an iterative algorithm for causal discovery with a human in the loop against benchmarks on the simulated and real datasets. Based on the methodology, an interactive web interface for doctors to input their expert knowledge and improve the learning results. If possible, you will assist in deriving the mathematical guarantee for the algorithm in convergence and stability. 4. Visualization and Presentation: You will prepare a manuscript for publication with a clear and engaging representation of the empirical results (optional).
- Gain a deep understanding of the health system and patient data gathering in a unique setting within the Swiss Paraplegic Center at Nottwil.
- Industry experience in data systems acquisition and processing through secure internal networks.
- Experience real data processing for training models.
- Testing in simulation and validating with real data.
- Learning and applying research methods Health Science.
- (Optional) Writing of a scientific publication.
- Gain a deep understanding of the health system and patient data gathering in a unique setting within the Swiss Paraplegic Center at Nottwil. - Industry experience in data systems acquisition and processing through secure internal networks. - Experience real data processing for training models. - Testing in simulation and validating with real data. - Learning and applying research methods Health Science. - (Optional) Writing of a scientific publication.
- Enrolled student at ETH Zurich or EPFL (or Swiss University)
- Strong statistic background
- Structured and reliable working style
- Strong programming skills in Python/Matlab
- Strong background in ML with experience in large data sets
- Understanding of explainable models (preferred) and sparse modelling
- Ability to work independently on a challenging topic
- Enrolled student at ETH Zurich or EPFL (or Swiss University) - Strong statistic background - Structured and reliable working style - Strong programming skills in Python/Matlab - Strong background in ML with experience in large data sets - Understanding of explainable models (preferred) and sparse modelling - Ability to work independently on a challenging topic
Host: Dr. Diego Paez (SCAI Lab).
Please send your CV and latest transcript of records from my studies to Dr. Diego Paez (diego.paez@hest.ethz.ch)
Host: Dr. Diego Paez (SCAI Lab). Please send your CV and latest transcript of records from my studies to Dr. Diego Paez (diego.paez@hest.ethz.ch)