Student projects

The following research topics are available for students that want to do their bachelor- or masterproject. A exact project should be defined based on the skills and the program of a student.

Developing an interface to stimulate protein intake of elderly

In the PROMISS project, the goal is to reduce malnutrition among elderly. Within this project, a prototype of a simple food intake measurement system should has been built. This involves physical computing devices (e.g. Arduino boards and sensors). In addition, a computational model of the cognitive and behavioural aspects related to food intake for elderly has been developed. This is based on existing models for behaviour change. The model will form the basis for a persuasive system that provides personalized guidance to elderly to increase their protein intake.

The goal of this project is to design and develop a user interface that communicates with the older people to stimulate them to adhere to their dietary advices, based on the measurements about food intake.

Review of Android Step Count Algorithms & Fitbit Step Accuracy for Physical Activity Tracking.

Goal: investigate to what extent built-in sensors correlate with Fitbit step counts. The aim is to compare this “in real life setting”, using different types of phones without precise instructions on how to handle the phone. Standard libraries (e.g. Google Fit, iOS Health) will be used and compared with the Fitbit. The outcome is an overview of the feasibility and accuracy of using built-in sensors for step count detection compared to the Fitbit activity tracker.

Technology is widely used for monitoring and supporting physical activity (PA), eg. via smartphone apps, wearables or activity trackers. Accelerometer-based physical activity apps are on the rise, moreover accelerometer-based activity trackers, like Fitbit are also conquering the market. Accelerometers emerge as the preferred sensors to measure PA under real-life settings, due to their non-obtrusive nature and low power consumption. However, obtaining accurate measurements of PA is a challenge due to the accelerometer sensor nature: data noise, sensor calibration, sensor quality, all influence the process.
There are numerous projects, in both research and industry settings, where complex algorithms are developed in order to obtain PA measures like number of steps, activity recognition. In this project, we are looking for students with interest in Android programming, lifestyle informatics, algorithms for physical activity, in order to:

  1. Conduct a systematic review of up-to-date open-source projects for step-counter or activity recognition, based on accelerometer sensor data.
  2. Perform real-life experiments with different types of smartphones to infer the feasibility and accuracy of the proposed open-source algorithms.
  3. Propose improvements for the existing ones, or develop new algorithms for PA tracking.
  4. Compare the accuracy of smartphone vs Fitbit activity tracker step counts.

Projects to stimulate physical activity

The aim of the Supreme Nudge project is to evaluate two types of interventions. One is addressing the environment in supermarkets and the other is addressing physical activity behaviour. For this latter part, an app will be developed that provides highly tailored and personalised feedback in coaching in order to make users more physically active. This app will build further on experiences from a previous project (Active2Gether), but needs to be adapted to the target population of Supreme Nudge, which are adults with a low socio-economic background. Furthermore, new technological and communication features need to be incorporated in the app. For that several sub-studies will be conducted, in order to test feasibility and potential effectiveness. This part of the Supreme Nudge project is a joint research project of the VU University (department of artificial intelligence?), the University of Amsterdam (department of communication science?) and Te Velde Research

Project 1: Social fencing

Goal: investigate how mobile phone sensors and / or IoT technology can be used for detecting social context . The ideas will be implemented and tested in a small pilot. The outcome is an overview of possible contexts that can be detected and an analysis of its feasibility.

Project 2: Geo fencing

Goal: investigate how mobile phone sensors and / or IoT technology can be used for detecting relevant physical contexts. This includes fencing significant indoor and outdoor locations with the goal of interacting with users based on their presence in these locations. The ideas will be implemented and tested in a small pilot. The outcome is an overview of possible contexts that can be detected and an analysis of its feasibility. This includes facets such as power consumption, accuracy of detection, and technical requirements.

Project 3: Detecting relevant activity

Goal: investigate how SWAN sensors and the outcomes of project 1 and 2 can be used for detecting relevant activities. The objective in particular is to detect sedentary and active behaviours, and to determine whether it is feasible to differentiate between different types of sedentary and active behaviour (i.e. between running, cycling, and swimming, or between watching TV and having coffee with a friend). Due to the power consuming nature of drawing on sensor information, special interest goes out to reliably detecting the above behaviours with as few sensors and as few timepoints as possible. The outcome is an overview of relevant activities that can be detected and an analysis of its feasibility, including information about power consumption, accuracy of detection and differentiation of activities, and technical requirements.

Determining trustworthiness of EMA ratings

In the E-Compared project, a mobile system has been developed that is used to monitor the mood of depressed patients during their therapy. This systems asks several times per day specific questions about a person’s mood, thoughts or feeling (EMA ratings). The goal of this student project is to investigate whether we can use detailed data about the way in which people answer those questions to determine the trustworthiness of the answers. One could think of aspects like “time between showing the question and the answer”, “time that finger pressed screen”, “physical distance of movement of finger”, etc. This project will be done in collaboration with the developers of the app, the University of Limerick. Not much programming is required.

Leave a Reply

Your email address will not be published. Required fields are marked *