Doing a MSc Thesis

This page contains description of how to do a MSc thesis with me at DTU. Before reading on, please read the official DTU Regulations on the MSc. Thesis. Here you can read the following:

“The objective of the master’s thesis is to give students the opportunity to apply the knowledge they have acquired in an independent way on a larger project that concludes with a written report. The thesis must document skills in applying scientific theories and methodologies to a clearly defined academic topic.” (my emphasis).

Then read Rasmus Paulsen’s guide on choosing the right project with the right supervisor.

Once you have read these pages, you may consult my list of project proposals for students to work with. All projects are somehow related to my research in personal health technology and you will hence be working together with a range of other researchers (PhD students) and programmers. Most of my research is centered within the following research areas:

When selecting a topic for your thesis, I recommend you to look at your previous courses and pick the topic in line with you favourite course(s). Note that a MSc Thesis is not a course – you don’t learn anything new – you apply what you already know to a specific project. Hence, if you want to excel in your thesis, you should pick a topic that you already  excel in.

Your background

Prospective students should:

  • Have a solid computer science background, previous exposure to the research areas listed above (HCI, UbiComp, mHealth) or related areas. This means that you probably is following a master degree at DTU Compute, such as:
  • I do consider outstanding students with other degrees (like Biomedical Engineering), but equivalent computer science background is required.
  • Have taken relevant DTU master courses on HCI, embedding programming, and software engineering, such as
  • Be a strong programmer and enjoy programming – most of my students’ research projects require very sophisticated programming, either because projects involve non-standard interfaces, or building underlying architectures.
  • Also be interested in non-technical cross-discipline aspects of HCI and health, such as sociological or psychological aspects of computer technology, design methods and thinking, usability & UX (graphical) design, health and disease management, and human physiology;
  • Be able to conduct user studies, including user-centered design, evaluations, controlled experiments, and pilot field trials, involving users such as patients, relatives, and clinicians
  • Team up as a pair – I alway recommend a MSc. thesis to be done in a team of no more, and no less than 2 persons. According to the DTU regulations, however, you can be up to 4 persons.

Supervision

My supervision is applying a group-based supervision approach. You can read more about this here. This basically means that:

  • There is a joint kick-off meeting for all students that initiates supervision.
  • Right after the kick-off meeting, we prepare the project database entry for DTU academic approval. This mainly states: title, description (10-20 linjes), starting date, ETCS points (normally 30), and ending date.
  • Then you start making the detailed project definition report including a plan for your work (see below). The has to be submitted to DTU within 4 weeks of the project. Once this is in place, you follow this plan (unless some radical problems occur).
  • All of my students meet bi-weekly for a joint supervision meeting. Before the meeting you prepare your progress report (see below). At the meeting you get supervision from me and you help each other. Each group have a 20 min. timeslot, so make the best out of it (i.e., be prepared).
  • You should work empirically (i.e., doing practical things like coding, analysing, running test, and evaluating applications) as well as theoretically (i.e., doing things like reading related literature, writing thesis chapters, and creating nice illustrations) throughout the entire project period. I DO NOT recommend writing the thesis during the last week of the project…
    • You may want to consult my “The Art of Doing a PhD” page – the description here applies equally for doing a thesis project, just in a smaller scale. Especially the Fish model can be useful.

  • I do not review your final thesis before you submit it. Supervision is a strange thing; on the one hand I am supposed to supervise you and yet on the other hand I’m also the one to review and grade your thesis in the end. Hence, in order to see what you can do independently and to grade your work, you need to write up your thesis on your own. This is the tail of the fish above.

The Project Definition Report

Besides the final thesis, the Project Definition Report (PDR) is the most important thing to write. The PDR contains the following outline:

  1. Project description
    1. Background
    2. Prior work
    3. Research question / hypothesis
    4. Contributions / goals
    5. Empirical / practical considerations
    6. Impact / innovation / application
  2. Intended learning objectives
  3. Plan
    1. Gantt chart
    2. Activities
    3. Milestones
    4. Deliverable, incl. chapters of the Thesis
    5. Risk analysis

Here is an example of a DTU Thesis Project Description Example, as it should look like. Here is a link to the template in Overleaf. On of the most important parts of the PDR is the Gantt chart, which is illustrated below.

According to the DTU regulations, the PDR should be made within the first month of the project. The PDR will be the major discussion item on the kick-off meeting.

Type of Thesis

In my experience, a MSc. thesis can be divided into five main types – all of which are related to the kind of ‘contribution‘ your thesis aims at:

  1. Theoretical – a theoretical thesis addresses a problem entirely theoretical, i.e. makes a contribution which is an abstraction of the problem. This can take many forms; a mathematical model (but then I’m probably not the right supervisor); a computational or simulation model; a conceptual model / framework; or a systematic literature review.
  2. Design – a design thesis focuses on a deep analysis of a particular domain and problem and through a thorough design process (typically with a deep involvement of users and stakeholder) come of with a system design. Here, focus is on your problem analysis and design skills, incl. user-centered design method, graphical user interface design, etc.
  3. Technology – a technology thesis focuses on the creation of a fundamental, general-purpose technology which can be used in several applications. This type of thesis appeals to most computer scientists and engineers and is one of the more typical ones I supervise. Here, focus is on your analytical and technical skills and you ability to create a novel and technically sound system design, and implement it. You should also evaluate your technology by creating (at least) two proff-of-concepts that demonstrate the generalizability of your solution. Examples include a programming framework for collection of activity data from wearable devices using a web-API and a software package for building cognitive tests for study apps on Android and iOS
  4. Application – in this type of thesis, you build an application, ie. a hardware / software system that ‘solves’ a problem. Like the technology thesis, this type of thesis also appeals to most computer scientists and engineers. Again, focus is on your analytical and technical skills and you ability to create a novel and technically sound system design, and implement a proof-of-concept of it. But in this type of thesis, focus is on creating an application which solves a real-world problem for real users/patients and hence involves user-centered analysis, design, and evaluation activities. Examples include designing a personal health technology for behaviour change in alcohol misuse or automatic fall detection using phone sensors.
  5. Study – in this type of thesis you take an existing technology (which you did not design yourself) and makes a thorough evaluation of it. Study theses falls normally into two broad categories; (i) technical study and (ii) user study – but when working in health technology, a third type exist; (iii) clinical study. In a technical study, you evaluate a system’s functional and non-functional software architecture qualities (e.g. scalability, security, etc.). In a user-study, you run a detailed usability study of the technology. In a clinical study, you run a study with patients to assess the feasibility of the system for clinical use (note that providing real-world clinical evidence will almost always be outside the scope of a MSc. thesis – this often require a PhD project).
  6. Dataset – this is a sub-type of the “Study” type of thesis where the purpose is to collect, structure, and make available a dataset. In this type of thesis, focus is on (i) creating a well-design data collection protocol, (ii) recruit a set study subjects, (iii) execute the study collecting the data, (iv) clean, structure, pre-process and document the data collected, and (v) make the dataset available (e.g., at DTU Data).

It is of crucial importance that you up front (in your PDR) decide which type of thesis you’re aiming at. I cannot emphasise this enough; I’ve seen a number of students struggling because they didn’t decide what they wanted to do with respect to the type of thesis, i.e. what the main contribution (aka ‘story’) of their thesis was. It is also important that you tell you supervisor what kind of thesis you’re doing – the choice of external examiner (Danish: ‘censor’) highly depends on this, and you don’t want your thesis being examined by the wrong type of examiner…!

Having said that; most theses contains elements of several types; it is hard to build a system without doing a design, and its a strong addition to a system thesis to also do an evaluation / study. However, the important part is here; what is your main problem statement and contribution.

For your inspiration, here are some MSc. theses, one of each type:

Theoretical

At the time of writing, I really don’t have a good example of a theoretical MSc. thesis.

Design

In 2018, Lys Egholm Andersen and Alina Gabriela Ciobanu did a thesis on “Echo: guided self awareness. Using personal data to support client self awareness through therapy“. This is a very good example of what I would call a ‘design thesis’ because it details the design of a specific app, including the design process and methods applied.  It also contains good examples of the different “design artifacts”, including the entire overview and micro-interactions of the final design.

A more recent (2020) example is the thesis “Breaking the cycle of Addiction by Design” by Margarita Genova and Nermen Ghoniem who did a very thorough design process and produced a set of impressive design artefacts.

Technology

Several of my MSc Thesis students contribute to the CACHET Research Platform (CARP) with different technology components, some of them being published as open source. Examples include:

  • Thomas N. Nilson (2020). “A Flutter Package for Real-Time Mobility Feature Computation”. [pdf] [GitHub] [pub.dev].
  • Mathias E. Boon (2020). “Removing Code Duplication Through Code Generation for Kotlin Web Services”. [pdf]
  • Richard Pekk (2022). “The Gardener Framework – An open-source programming framework for collection of wearable activity and health data from web-based services”. [pdf] [Github].

Application

In 2008, Niels Nørskov did a thesis on “Improving Patient Safety in the Operating Room Using Context-Aware Technologies and RFID” which resulted in the design of the “Context Aware Patient Safety and Information System” (CAPSIS). This was later turned into a scientific publication [1].

A more recent DTU example is that Andrea Quemada did a MSc thesis on “Personal Health Technology for Behaviour Change in Alcohol Misuse“.

Study

In 2008, Sanne Jensen, Joan Nielsen Meyer, Jesper Dahl and Annabel Lee Krarup did a study of the CAPSIS system. This study was done in a simulated clinical environment at the IT Experimatarium at Herlev Hospital. The thesis – entitled “AXHIS: Metode til vurdering af it- systemer inden implementering” (in Danish) provided both a thorough evaluation of the system as well as a proposal for a systematics clinical simulation methodology. Later, Sanne Jensen continued this work in her PhD Thesis [2].

Dataset

I don’t have an example of a MSc Thesis creating a dataset. But we did create a dataset as part of the REAFEL / mCardia project, which is published at DTU Data and documented i a paper [3], which to a large degree resembles what a MSc Thesis on a dataset would look like.

Supervision Meetings – the PPP Model

As said above, I always run group-based supervision. This has a lot of benefits, not least that you are able to help each other. I expect all my students to be open-minded and helpfull towards each other. Some are good at solving technical problems, others are good at LaTex, and others are good in writing. So – please help each other out.

In the meetings we use the PPP model, where each group/student present:

  • Progress – what have you done since the last meeting. You should prepare and show tangible artefacts that you have worked on. This can be the project plan, a literature review, some UX design, a software architecture, code, a running demo, or even some hardware setup.
  • Problems – what problems, challenges, or questions do you have. What do you want to have my – and the rest of the group’s – input on?
  • Plan – what do you plan to work on for the next 2 week period. Do you need me or others in the time in-between supervision meetings?  Note that I run a pretty tight schedule, so it is important to book me for supervision outside the bi-weekly slots.

Writing the Thesis

According to the DTU regulations for the MSc Thesis, the thesis must be written in English (otherwise, an approval from the Head of Studies must be obtained).

A Thesis is structured according to the type of thesis. However, chapter 1 is almost always structured in the same way, as described by Saul Greenberg on his “Deconstruction Chapter 1” page.

As Saul Greenberg puts it:

“Almost all chapter 1’s contain the following structure. You can use this as a ‘formula’ to create your own draft of Chapter 1. Of course, you can alter and deviate from this formula, but only do so if you have a good reason for it.”

Please always follow this outline.

Chapter 2 is most alway a literature review of related work (the chapter is often entitled “Related Work”). Again, Saul Greenberg has a good description of how a literature review can be done. Note, however, that this guide is targeted PhD students, and as a MSc student, you should only include a lighter version of a literature review.

I often get the question of whether commercial systems should be included in the “Related Work” chapter. This is a tricky question, and the answer is “if relevant”. For example, if you’re doing a thesis on a particular type of technology, there might be commercial technologies that are relevant to introduce, and discuss what you’re doing is similar/different/related. But remember that it should be commercial technologies, which are relevant to what you are doing, i.e. technologies that address the same research question. It is not relevant to discuss commercial technologies that you have used to build your own system (e.g., docker, kafka, spring, …). These are “merely” tools or implementation details, which are introduced later in the thesis.

Patents are very rarely included in chapter two.

Length of the Thesis

I’m often asked what should the length of the thesis be? Well – there are no formal requirements. However, the following guidelines applies:

  • You should use the DTU Latex template (see below) which sets the standard for font size, etc.
  • A typical thesis is 40-70 pages long for the main sections, excluding pre-matter (summary, foreword, table of content) and appendices.
  • A thesis must NEVER exceed 100 pages.

You should make clever use of appendicies where you can put details that you want to documents, but which are less important to the main story of the thesis. This includes detailed design diagrams, interview transcripts, pictures, detailed statistics (including graphs), etc.

Stating “Distribution of Contributions”

According to National Danish regulations, grades are given on an individual basis. Therefore; if you are doing a MSc. Thesis as a group (which I, as mentioned above, recommend), it is important that it is clear which member of the group did what. There are two ways to do this:

  1. Each chapter has a separate author. In this case, the thesis is like an edited book, with each chapter written by an individual person. And of course, the person who wrote the chapter also did the work/research described in this chapter.
  2. You include the mandatory Appendix A entitled “Distribution of Contributions” which in details outline for each group member what s/he did – both in terms of the empirical work (e.g. interviews, transcribing, coding, testing, evaluation) and in terms of writing the thesis.

It is NOT permitted to write something like “The work was done equally by all members of the group”.

Note that if the distribution of contributions are not explained, we cannot evaluate the thesis and hence grade you!

Hand-in

The main hand-in is clearly the Thesis, as explained above.

However, you should alway also hand-in – or make available for download – the empirical material. Examples of empirical material are:

  • Source code – can be made available e.g., on Github or as a zip file.
  • Binary, executable application – which can be installed and run “out-of-the-box”.
  • Dataset – a data set collected as part of a study.
  • Analysis scripts – like Jupyter notebook or MatLab scripts
  • Design artefacts – like physical mockups, pictures, graphics, and mock-ups (e.g., in Figma)
  • Video – a video (e.g., on Vimeo) that shows a demonstration of an application or the execution of a study (e.g., collection of data from participants)

Note that it is important to nicely structure and document this empirical material. You should add README files to Github repros, make API documentation, make documentation of analysis scripts, put you design artefacts in a nice “portefolio”, and make prober speech and/or text in the video(s).

A big unstructured zip file with stuff you have on you harddrive is NOT a good submission.

Oral Exam

Sune Lehmann have provided a nice description of how the MSc. Thesis oral exam runs. Please read this.

I have only one addition; in line with the description above, we grade you on an individual basis. This implies that each member of a group will have his/her own individual oral exam. This is also stipulated in the DTU regulations for the MSc Thesis.

Practical Things

  • We typically use Slack for coordination and communication. Each project has a separate channel.

LaTex

It is mandatory to write your thesis, PDR, and progress reports  in LaTex. There is plenty of support for LaTex at DTU, including the Thesis template

References

[1] [pdf] J. E. Bardram and N. Norskov, “A Context-aware Patient Safety System for the Operating Room,” in Proceedings of Ubicomp 2008: Ubiquitous Computing, 2008, pp. 272-281.
[Bibtex]
@inproceedings{ubicomp2008:bardram,
Author = {Jakob E. Bardram and Niels Norskov},
Booktitle = {Proceedings of Ubicomp 2008: Ubiquitous Computing},
Pages = {272-281},
Publisher = {ACM},
Title = {{A Context-aware Patient Safety System for the Operating Room}},
Url = {http://portal.acm.org/citation.cfm?id=1409672},
Year = 2008,
}
[2] [pdf] S. Jensen, “Use of Clinical Simulation in Development of Clinical Information Systems,” PhD Thesis, 2014.
[Bibtex]
@phdthesis{jensen2014use,
title={Use of Clinical Simulation in Development of Clinical Information Systems},
school = {Aalborg University, Denmark},
author={Jensen, Sanne},
year={2014}
}
[3] [pdf] D. Kumar, S. Puthusserypady, H. Dominguez, K. Sharma, and J. E. Bardram, “CACHET-CADB: A Contextualized Ambulatory Electrocardiography Arrhythmia Dataset,” Frontiers in Cardiovascular Medicine, p. 1702, 2022.
[Bibtex]
@article{kumar2022cachet,
title={CACHET-CADB: A Contextualized Ambulatory Electrocardiography Arrhythmia Dataset},
author={Kumar, Devender and Puthusserypady, Sadasivan and Dominguez, Helena and Sharma, Kamal and Bardram, Jakob E},
journal={Frontiers in Cardiovascular Medicine},
pages={1702},
year={2022},
publisher={Frontiers}
}