Postdoc Researcher (2 Years) - T150-01201 BE-FIT SEC

For the purposes of the project, Singapore is an ideal choice. It’s population is highly tech-savvy, its healthcare system is clearly structured and there is a critical mass of accessible patients. Critically, Singapore is (and will continue to) face one of the largest increases in the proportion of elderly in its population. It is likely that Singapore will rank among the top 10 “oldest countries” together with other Asian and European nations.

Singapore also happens to be one of the best places to live in Asia. HSBC’s annual survey rates it as the best cities in the world to live for expats, while Mercer’s rates it to have the best quality of life in Asia. The reasons are many, but primary factors are efficient public transport, and education systems and substantial health care industry. It is also a very clean and safe city.

In collaboration with the Woodlands Health (WH) and the National Health Group (NHG), the Rehabilitation Research Institute of Singapore (RRIS), the Nanyang Technological University (NTU), and SingHealth, SEC is undertaking a research programme on "Built Environment and physical activity in Falls and Arthritis study (BE-FIT)". It addresses imminent health challenge on moving away from “sickcare” and pivoting towards preventive healthcare as part of the nationwide effort on Healthier SG. Within BE-FIT we envision motivating vulnerable older adults to engage in healthy behavior by providing recommendations on improving accessibility (as well as preception thereof), in urban environment for uptake of physical activity. A deeper understanding of the interactions and interplay between Built Environment (BE) and the high burden of falls and osteo-arthritis (OA) as proposed within the BE-FIT is crucial towards informing data-driven decisions on urban design and how the mobility-impaired elderly interact with their physical environment. Within the BE-FIT framework we are advertising the job position for a postdoctoral researcher.

Falls result in severe physical as well as psychological impact among older adults. Beyond physical implications on injury-related trauma and in severe cases death, the psychosocial impact of falling can also be excruciating. Fear of falling can result in vicious cycles of decreased activity as well social isolation. These in turn lead to lower muscle strength and higher risk of future falls. In a similar manner, osteo-arthritis (OA) can lead to fear of movement (kinesiophobia) resulting in reduction of physical activity levels.

We will investigate movement patterns and features of walking outdoors and in the neighbourhoods among vulnerable older adults (suffering from OA as well as at high risk of falling) in order to understand perceptions on interacting with built environment. We will acquire these movement patterns and features using the state-of-the-art inertial measurement units (wearables such as ZurichMOVE or Axivity) sensors. These sensors are equipped with triaxial accelerometers and gyrospcopes and provide assessment of aspects such as impact and swing behaviour during different movements. Specifically we will be addressing the following research questions:

1. What are the kinematic characteristics of walking among older adults with OA and/or previous falls under ecological settings (neighbourhoods)?
2. Do the kinematic characteristics of naturalistic walking predict physical activity rates?
3. Does the design of walkways (including overall layout, e.g. design of curbs, pathways etc and accessibility features e.g. size of the curbs, or height of side walk, ramps vs stairs) impact overall levels of physical activity as well as specifics of walking quality?

The general concept is to generate machine learning as well as statistics-based models to extract movement patterns from movement/locomotion/walking/gait dataset collected via wearable sensors. The gait data from multiple sensors will be collected while participants (older adults) move/walk/transition for both short (up to 10 minutes) as well as long (over multiple days) periods of time. This data will be complemented with data from our collaborators on walkability assessment, where we evaluate naturalistic observation on individual’s daily routes and outdoor activities. Finally, we will also tap into questionnaire data involving individuals’ fall history, psychosocial status as well as cognitive ability, as well as their perception of the built environment. The primary task will be to extract features (gait signatures, but also artificial “machine learned” features) that allow us to assess fall risk in an individualized manner. Crucial aspects are the interpretability and repeatability of these signatures as these aspects will allow clinical as well as stakeholder uptake. Another important aspect for uptake is the association (via analysis as well as interpretation) of these features to the walkability as well as clinical assessment to provide a hybrid ‘mapping’ of the manner in which older adults interact with their environment.

Minimum 2 years of experience in machine or deep learning with a background (PhD) in computer science, computer vision, neuroscience, physics or biomedical and/or other engineering fields. Expertise in predictive model development, especially for healthcare applications. A solid understanding in experimental design, feature extraction, selection, and analysis, as well as tailoring machine/deep learning techniques to hybrid datasets including clinical battery, and objective physiological (movement) datasets. A strong foundation in deep and machine learning algorithms, statistical analysis, and study design from ideation to evaluation and validation. A strong publication record especially in area of artificial intelligence and machine learning. Presentations at conferences and participation in workshops is desired.

Programming skills:

• Expert in Python (PyTorch as well as the use of libraries).
• Experience/expertise in working with Matlab and/or R (or any other statistical software e.g. SAS).
• Previous experience with physiological (such as heart rate via an ECG or also EEG) datasets is desired, but not a must-have.
• A solid foundation of machine learning frameworks such as Tensorflow and algorithms, statistical analysis, and study design from ideation to evaluation and validation.
• Experience with GUI development is desired.

Personal:

• Are you motivated to work on challenging problems?
• Can you work independently on a project level demonstrating problem solving skills?
• Do you see yourself fitting in with the team of multinational group of biomechamists, engineers as well as health-care and clinical scientists?
• Do you have a penchant for collaborating - maintaining channels of communication - with lab/team members but also worldwide? If yes, this job might just be for you.

The Singapore-ETH-Centre is an equal opportunity and family-friendly employer. All candidates will be evaluated on their merits and qualifications, without regards to gender, race, age or religion. The employment will be at the Singapore-ETH Centre and local working regulations will apply.