Project Guidelines

The following guidelines are designed for 6-week research projects.

Overview

I tend to have three aims when supervising students on research projects:

To provide students with experience at each stage of a research project.
To equip students with essential skills for conducting scientific research.
To provide opportunity to perform novel research which could impact healthcare.

Typically, students are asked to complete the following tasks:

Week	Tasks
Pre-project	Pre-reading
1	Learning objectives Literature review Research Plan Data Access
2	Literature review Research Plan Data Access
3	Experiments Project Report
4	Experiments Project Report
5	Experiments Project Report
6	Project Report Project Presentation
Post-project	Conference Submission

In my experience, 6 weeks is an appropriate amount of time in which students can gain experience at each stage of a research project, develop new skills, and make valuable contributions to our ongoing research studies.

Pre-reading

Pre-project

You may want to start work on the project early, in which case I would recommend reading some key articles in the field. These could be on topics such as:

The clinical background to the project.
Current approaches being used to tackle the problem.
Previous research conducted by our group which relates to the project.

I will be happy to recommend some suitable articles.

Learning objectives

Week 1

It is good to firstly think about what you would like to learn from the project. You may find it helpful to list some learning objectives, which could include:

Being able to explain the clinical background to the problem
Gaining experience across each of the different stages of a research project
Developing particular research skills (e.g. writing a literature review, using statistical analysis techniques, writing code)
Preparing an abstract suitable for conference submission
Understanding good practice for clinical research (such as by undertaking a Good Clinical Practice course)

We will discuss your hopes for the project during the first week. This process will help me tailor the project towards your personal learning objectives.

Literature Review

Weeks 1-2

A literature review is a key step in performing valuable scientific research. It allows you to: (i) understand the problem being tackled; (ii) appreciate current approaches to tackle the problem; and (iii) identify potential areas for further research to move the field forward. Your literature review will provide a basis with which to define the aim(s) of your project, and potentially provide insight into how you could design your experiments to address these aims (perhaps building on the methodology used in previous studies). Therefore, it is important to get started on a literature review early in the project, preferably performing most of the work in Week 1.

I would suggest the following steps to create a literature review:

Explore your project: Discuss the project with me, and have a quick look into key topics underpinning the project, in order to understand it more fully. Explore our current understanding of the three items listed above: (i) the problem being tackled; (ii) current approaches to tackle the problem; and (iii) potential areas for further research.
Search for relevant publications: Use search engines to identify key publications relating to the project, bearing in mind the three items above. You will probably find too many publications to read in a year, let alone a 6-week project, so be selective. Often a review article can help to understand the broader context in which the literature sits - feel free to ask me if I would recommend any particular review articles.
Extract information from the publications: Having identified key publications, read them and make notes on any relevant points in the publications. Relevant points could include: (i) the clinical significance of the problem being tackled; (ii) the aims and findings of previous research in the field; (iii) the methodology used in previous studies; and (iv) suggested directions for future research (which could either be stated explicitly, or which you may identify by recognising gaps in the research conducted).
Sort the information: Categorise and sort the information into themes - ideally you should expect to write one paragraph on each theme. Remove any information which you think isn’t too relevant.
Structure your writing: You are now in a position to structure your literature review. I would suggest an Introduction and a Conclusion, and between these, a paragraph on each theme. The paragraphs on each theme could follow this format:
- Point: A single sentence summarising the point you will make (e.g. Most of the literature on screening for atrial fibrillation has used 12-lead ECG recordings.)
- Quotations: Several sentences in which you refer to the evidence supporting this point - perhaps one or two sentences per publication (e.g. Doliwa et al. assessed the performance of a single-lead ECG device against reference diagnoses from 12-lead recordings [ref].).
- Comment: A single-sentence providing your (critical) view on the significance of this point. (e.g. 12-lead ECGs are often viewed as a gold standard reference against which to compare single-lead ECGs, as demonstrated by the inclusion of such studies in NICE Guidance relating to Lead-I ECG devices.)
Write your review: Perhaps start by writing the paragraph on each theme, and then write the Introduction and Conclusion last.
Provide references: Use a reference manager to provide inline citations, and a list of references (feel free to ask me for recommendations).
Proof-read your review: Ideally, don’t look at the review for a couple of days after writing the first draft, and then re-read it to identify any areas for improvement.

Further thoughts: Ideally, a literature review should critically appraise the literature, rather than simply stating what has been said in the literature. Usually, a review of between 1,000 and 3,000 words is appropriate. You may wish to add to your literature review during the project if you find additional relevant publications.

Research Plan

Weeks 1-2

The next key step in conducting scientific research is to make a research plan. This will summarise the motivation for your project, its aim(s), the methods you intend to use in your experiments, and the expected outcomes of the project. A well-written research plan can make it much more straightforward to conduct experiments and write them up.

I would suggest trying to summarise the following on a single side of A4:

The clinical background to the project: 1- 2 sentences
The problem being tackled: 1- 2 sentences
Current approaches to tackling this problem: 1- 2 sentences
The aim(s) of this project: 1- 2 sentences, indicating how your project will help tackle the problem. You could have an overall aim, and then a few (perhaps 3) objectives, where each objective refers to a research question which you intend to answer.
The methods to be used in this project: 4-5 sentences providing key details of the methods you intend to use, such as the dataset and a brief overview of the analyses you will conduct. You may find it helpful to split the methods into parts (e.g. one part per research question).
The expected outcomes: 1- 2 sentences on how this project will help tackle the problem.

The research plan does not need to be detailed - for instance, you probably won’t know at this stage how many participants are in the dataset, or which statistical tests you might perform. Rather, it should provide a high-level summary of the project.

Data Access

Weeks 1-2

It is likely that you will use one or more datasets in your project. These could be publicly available datasets, or private datasets hosted within an institution:

Publicly available datasets: If you are using a publicly available dataset (such as those listed here), then hopefully it should be relatively straightforward to access the data. I have provided some MATLAB code to help collate some of these datasets into suitable formats for analysis here.
Private datasets: If you are using a private dataset (such as the SAFER dataset) then you may need remote access to the data. Discuss this with me, and if you do require remote access to a dataset hosted in Cambridge, then it is likely you will need to follow the instructions provided by the Clinical School Computing Service (here) in order to gain remote access. Specifically, you will need:
- A suitable login: If you are using a temporary account then you will need to reset its password before using it here.
- A VPN connection: You will need to connect to a VPN in order to have permission to access the system. See the instructions here on setting up a VPN. Note that you will be using the settings for the Biomedical Campus. If you have been given a temporary login then you will use that to login to the VPN.
- Remote desktop software: If you are using a Windows computer at home, then see these instructions on how to remotely connect to the Cambridge computer from your home computer. Ask me for the name of the Cambridge computer that you are connecting to, and if you have been given a temporary login then use that to login to the computer. If you are using a Mac at home, then you will need to use a different piece of Remote desktop connection software - ‘Microsoft Remote Desktop’, which is free in the App Store.

Experiments

Weeks 3-5

We will discuss your planned experiments when discussing your Research Plan. It is likely that you will need to obtain new skills and develop new skills to carry out the experiments. Here are a few general hints, although much of the experimental methodology will be particular to your project, so is best discussed between us.

Answering research questions: Your experiment(s) should aim to answer one or more research questions (as listed in your Research Plan). In general, it is better to answer one question well, than to answer several questions less well. The remaining hints are designed to help you answer research questions well.
Understanding the dataset: It will take time to understand the dataset(s) you are using. This is an important part of the project, and investing time at this stage will make conducting the experiments much easier.
Designing appropriate analyses: Once you have gained an understanding of the dataset, you will be in a position
Curating the dataset: You will probably only need particular variables from the dataset for your research. You may also only need data from particular subjects who meet inclusion criteria. Furthermore, you may need to derive new variables from those in the dataset. I recommend taking time to curate the dataset to prepare it for the analyses. Document how you curated the dataset, and understand any assumptions made and their potential implications on the results (such as the criteria used for identifying subjects for inclusion, which may limit the generalisability of the conclusions to only subjects meeting those criteria).
Performing analyses: You will need to use statistical software such as Excel, MATLAB, GraphPad, or others, to perform the analyses. I recommend that you make the analyses as reproducible as possible. Firstly, make sure you keep a copy of the original data files, so that you can always get back to the original data. Secondly, if you know how to code, then consider creating a script to perform the analyses. This makes it easy to modify the methods you use if you decide to change them during the project.
Documenting analyses: Keep a record of your experiments: the methods you use, and the results you obtain. This can take many forms, such as paper notes, a logbook (the traditional way), or electronic notes (I use Joplin). The objective here is to be able to recall what you did, so that you can accurately document your methods in your project report. To this end, you may also find it helpful to write bullet point notes in your Project Report as you are conducting the experiment, to help you when writing up the report.
Discussing analyses: We will aim to meet at least once per week during the project. It would be helpful if you could summarise the work you have done on the analyses during these meetings. This will help me to provide specific guidance on your work.

Project Report

Weeks 3-6

Your project report will probably be the most important output from your project. It will enable others to learn about your work and make use of it, whether in research or clinical practice. I’d encourage you to structure your report using the standard structure of articles reporting scientific experiments (with some additional flexibility):

Abstract: (write this last) An Abstract provides a concise overview of a scientific study. Typically 200-300 words in length, it should give the reader an overview of the study, containing only the most important and essential points. The reader should be able to decide, based on the Abstract, whether they want to read the accompanying paper (or, in this case, the remainder of the Project Report). Further details on writing an abstract are provided below (see the Section on Conference Submission).
Introduction: The introduction should:
1. Motivate the work: Provide the motivation for both the broad topic of your work (e.g. identifying atrial fibrillation) and the specific issue that you have focused on (e.g. accurately identifying ECG recordings for clinical review). Minimum length: 1 paragraph.
2. Summarise the state-of-the-art: Describe the state-of-the-art in tackling the specific issue, making reference to key publications (perhaps 5-10 references). If there really is no prior art in tackling this particular issue, then refer to works which have tackled similar issues. Minimum length: 1 paragraph.
3. Introduce your study: State the aim of the study in a single sentence. If necessary, use an additional sentence to state the objectives (i.e. the smaller aims). Describe key aspects of the methodology used in your experiments in 1-2 sentences, focusing on those which make your experiment(s) particularly valuable (e.g. using a large dataset which is representative of real-world conditions). Whilst there is no need to state the results at this stage, you could briefly mention how the results are significant (e.g. “the results will be valuable for the design of AF screening programmes”). Minimum length: 1 paragraph.
Literature review (optional): Although this section isn’t usually included in a scientific article, it is likely that your literature review was a key component of your project. Therefore, you may wish to include an additional section in the report in which either the literature review is inserted in full, or a summary of it is inserted. Examples of articles in which an additional ‘literature review’ section was inserted after the Introduction are this one (in which the section is called “Respiratory rate algorithms”), and this one (in which the section is called “Review of previous work”).
Methods: In this important section you should provide a comprehensive description of the methods used in your experiment(s). Ideally, the description should be sufficient for an experienced researcher to be able to repeat your experiments (a bit like a cooking recipe). It is helpful to split the ‘Methods’ section into subsections, which could include:
- Dataset(s): A description of the dataset(s) used in the experiments. This should provide the key information about aspects of the dataset which relate to the experiments. This will help the reader understand why the dataset was suitable for addressing the research question(s). You do not need to provide a comprehensive description of the study in which the dataset was originally collected - only a brief summary, and those details which are relevant to your study need to be included.
- Data processing: This should describe how the data were processed in order to perform the experiments. For instance, you may have only used data from individuals meeting specific criteria (such as those aged over 75), or you may have calculated new variables from those in the dataset (such as calculating the Body Mass Index from height and weight). These methods should be described.
- Statistical analysis: This section should state any statistical techniques or tests you used (e.g. Pearson’s correlation coefficient), and what they were used for (e.g. was used to assess the correlation between X and Y). There is a tension between providing all the necessary details, and keeping the description concise. You will need to use your judgement to discern which details are required, and which should not be included, perhaps because: (i) they are intuitive to someone experienced in this field of research; or (ii) they had little impact on the results.
Results: This section should present the results of the experiments. Key results are most often presented in Figures and/or Tables, which can help to quickly convey them to the reader. Key results are also stated in the text, e.g. “as shown in the box plot in Figure 1, there was a significant difference between the heights of 5 and 15 year olds (100 vs. 150 cm, p<0.05)”. Minor results are usually provided in the text. It is useful to state the key findings in the text (for instance, the significant difference in heights is stated here). However, further discussion of the results should be reserved for the ‘Discussion’ section.
Discussion: Discuss the key findings and their implications. It is helpful to start with an introductory paragraph stating the key findings and the key aspects of the experimental methods which led to these findings. This paragraph should emphasise the novel points of this study. The remainder of the discussion should consider:
- Context: How the results sit in the context of the wider literature. What was known before this study, and how does this study add to our understanding?
- Significance: What are the implications of the results? For instance, could they impact future research, device design, or clinical practice?
- Limitations (often a subsection): What were the limitations to this study? How generalisable are the results?
- Future work (often a subsection): Discuss potential directions for future work to build on this study.
Conclusion: A short section stating the key conclusion(s) of the work. Often this only needs to be a few sentences in length.
Acknowledgment: Acknowledge any funding providers.
Appendices (optional): In many reports there is no need for appendices. However, they can be a useful way of documenting additional work you have done (such as results which don’t relate to the main research question(s)).

You may find it helpful to use a template for your report, such as the Microsoft Word or Latex templates provided in the ‘Instructions for Authors’ section of this webpage.

Project Presentation

Week 6

Conference Submission

Post-project

Requirements for conference submissions vary between conferences, and can include submitting an abstract, a 1-page overview, or a short paper. Most conferences require an abstract, either as the sole requirement, or as part of a short paper.

Writing an abstract

I suggest writing the abstract after finishing your Project Report, as it should summarise the entire study. An abstract can typically be structured to contain a few sentences on each of the following:

Introduction: Provide essential background to motivate the study, and state the study’s aim. For instance, here’s a suggested template for writing the introduction (i.e. first part) of an abstract: (based on this article, reproduced under CC BY 4.0)
- State the big problem, e.g. “Atrial fibrillation (AF) is a common irregular heart rhythm associated with a five-fold increase in stroke risk.”
- State the smaller problem being tackled in this study,, e.g. “It is often not recognised as it can occur intermittently and without symptoms. A promising approach to detect AF is to use a handheld electrocardiogram (ECG) sensor for screening. However, the ECG recordings must be manually reviewed, which is time-consuming and costly.”
- State the aim(s) of the study, e.g. “Our aims were to: (i) evaluate the manual review workload; and (ii) evaluate strategies to reduce the workload.”
Methods: A few sentences summarising the key aspects of the methods, which are essential to understand the study. For instance, it could include “Participants used an ECG recorder four times per day for three weeks. The ECG signals were post-processed to identify heartbeats, calculate inter-beat intervals, and detect AF based on a Poincare analysis” . It would be less likely to state the exact device used to record the ECG signals, or the exact algorithm used to identify individual heartbeats, as these are not likely to be required to give an overall understanding of the study. The reader can always read the full paper if they want to find out specific details.
Results: State the key results which informed the conclusion(s).
Conclusion: State the main conclusion(s) in one or two sentences. Optionally, also state the significance of these conclusions. e.g. “In conclusion, screening for AF using handheld ECG devices requires an average of 35 ECGs to be manually reviewed per AF diagnosis. Our new approach reduces this workload to 20 ECGs per AF diagnosis, with minimal reduction in sensitivity. This approach could be incorporated into existing screening approaches to reduce workload and thereby increase the cost-effectiveness of screening.”