added robots and safer projects pages

robots.txt

@@ -0,0 +1,2 @@
+User-agent: *
+Disallow: /safer_projects/

safer_projects/introduction.md

@@ -0,0 +1,29 @@
+---
+layout: default
+title: 1 Introduction
+parent: SAFER Projects
+has_children: false
+---
+
+# Introduction
+{: .no_toc }
+
+This page provides an introduction to our work on the SAFER Programme.
+{: .fs-6 .fw-300 }
+
+## Table of contents
+{: .no_toc .text-delta }
+
+1. TOC
+{:toc}
+
+---
+
+## The SAFER Programme
+
+The [SAFER Programme](https://www.safer.phpc.cam.ac.uk/) is nvestigating whether using ECG devices to screen for atrial fibrillation (AF) is effective and cost-effective in reducing the incidence of stroke. It is being conducted by researchers in the Department of Public Health and Primary Care at the University of Cambridge. Many thousands of participants have taken part in the research programme.
+
+The Programme consists of three phases:
+- The Feasibility Study
+- The Remote Feasibility Study
+- The Trial

safer_projects/reading/analysis.md

@@ -0,0 +1,12 @@
+---
+layout: default
+title: Analysis
+parent: SAFER Wearables
+has_children: true
+---
+
+# Analysis
+{: .no_toc }
+
+This page provides links to analysis tools being used in the SAFER Wearables Study.
+{: .fs-6 .fw-300 }

safer_projects/reading/data_curation.md

@@ -0,0 +1,12 @@
+---
+layout: default
+title: Data Curation
+parent: Analysis
+grand_parent: SAFER Wearables
+---
+
+# Data Curation
+{: .no_toc }
+
+... .
+{: .fs-6 .fw-300 }

safer_projects/reading_material.md

@@ -0,0 +1,120 @@
+---
+layout: default
+title: 2 Reading Material
+parent: SAFER Projects
+has_children: false
+---
+
+# Reading Material
+{: .no_toc }
+
+This page provides a background reading material relating to our work on the SAFER Programme.
+{: .fs-6 .fw-300 }
+
+## Table of contents
+{: .no_toc .text-delta }
+
+1. TOC
+{:toc}
+
+---
+
+## Initial Reading for SAFER Student Projects
+
+This page some recommended reading for projects within the SAFER Programme. Do be selective about which articles to read - there is certainly no need to read them all. In addition, be selective about which parts of the articles to read - for instance, simply reading the abstract may be sufficient in some cases.
+
+**PDFs** of documents which are not available online are provided [here](https://universityofcambridgecloud-my.sharepoint.com/:f:/r/personal/pc657_cam_ac_uk/Documents/SAFER_Engineering_Resources/SAFER_student_projects/Project_Resources/Reading_Materials). _NB: to open links on this page, right click and open in a new tab._
+
+---
+
+## Atrial Fibrillation
+
+The following are recommended for initial reading on the atrial fibrillation (AF):
+
+- **AF is associated with a fivefold increase in stroke risk:** See [this paper][2] which demonstrates the importance of AF.
+- **AF is responsible for approximately 28% of all strokes**: See [this paper][4]
+- **Approximately 3.3% of the UK population have AF:** See [this paper][3]
+- **The risk of stroke can be reduced by approximately 60% through anticoagulation**: See [this paper][5]
+- **AF is often unrecognised:** See [this report][1] which suggests that there may be approximately 425,000 people in England with undiagnosed AF.
+- **The global prevalence of AF is expected to more than double by 2050:** See [this paper][6]
+- **AF is diagnosed using an ECG assessment:** Section 3.2 of the recent [guidelines][9] on AF provide details of the criteria used to diagnose AF from an ECG. (there's no need to read the rest of the guidelines)
+
+
+## Screening for Atrial Fibrillation
+
+The following are recommended for initial reading on screening for AF:
+
+- **An introduction to screening for AF:** [This review][10] provides an accessible introduction to screening for AF.
+- **The SAFER Programme:** Try to learn about the SAFER Programme, using the links [here](https://peterhcharlton.github.io/info/tools/reading/af_screening.html#safer-study). You will likely use data from the SAFER Feasibility Study in your project - [this publication](7) describes the dataset (although note that the number of AF diagnoses increased slightly after this paper was written), and [this paper](http://peterhcharlton.github.io/publication/prioritising_ecgs/) contains an initial analysis of the ECG data.
+- **The STROKESTOP Study:** The STROKESTOP Study used very similar methodology to the SAFER Programme. There are several links on the study [here](https://peterhcharlton.github.io/info/tools/reading/af_screening.html#strokestop-study). However, I would recommend just reading [this paper][8].
+- **The ECG recording device:** Have a look at Zenicor's device [here](https://peterhcharlton.github.io/info/tools/reading/af_screening.html#zenicors-handheld-ecg-device) to get an idea of the device used in these studies.
+
+
+## Additional Topics
+
+Whilst the above topics are likely to be of some relevance, the following topics may or may not be relevant. Do be selective.
+
+### ECG signals
+
+- **[The physiological origins of ECG signals](https://www.mit.edu/~gari/ecgbook/ch1.pdf):** provides an introduction to what an ECG signal is. It's important to have some understand of what P, Q, R, S and T-waves are.
+
+### ECG Interpretation
+
+- **[2020 ESC Guidelines for the diagnosis and management of atrial fibrillation][28]:** This is a lengthy document, so please don't read it all. However, Section 3.2 (and the Table called 'Recommendations for diagnosis of AF' are particularly helpful). It provides details of the standard criteria for identifying AF in single-lead ECGs (such as those collected in the SAFER Programme).
+
+### ECG Analysis
+
+- **Traditional ECG signal processing:** [This book chapter](http://diec.unizar.es/~laguna/personal/publicaciones/libroWiley.pdf) provides a helpful overview of traditional ECG signal processing techniques. [This article](https://doi.org/10.1109/ACCESS.2020.3026968) is also helpful.
+- **The Cardiolund algorithm:** For details of the automated algorithm used to analyse ECGs in the SAFER (and STROKESTOP) Programmes, see [here](https://peterhcharlton.github.io/info/tools/reading/af_screening.html#cardiolunds-ecg-parser-algorithm). Other companies also produce similar ECG analysis algorithms (e.g. [Cardiomatics](https://cardiomatics.com/)), although it is probably most important to read about the Cardiolund algorithm as it is being used in SAFER, and is also well-reported in the literature.
+- **ECG beat detection:** [This page][27] provides a list of open-source beat detectors, which may well be helpful if you intend to analyse the ECG signals yourself (rather than simply using the outputs of the Cardiolund algorithm).
+- **Heart rate variability analysis:** The following paper provides a useful introduction to heart rate variability (HRV) analysis: [An overview of heart rate variability metrics and norms][29], and [this paper][31] provides additional details.
+- **Details of ECG signal processing:** For a comprehensive overview of signal processing techniques used to predict AF from the ECG signal, see [this thesis](http://hdl.handle.net/10362/64177), focusing on Sections 5.2.2 and 5.2.3.
+- **ECG signal filtering:** [This book chapter][11] and [this one][12] are written by experts in the field.
+- **Noise in an ECG signal:** See the first part of [this article][13]
+- **Deep learning for the ECG:** [This][20] is a very well written paper on using deep learning with the ECG. See [this paper][19] which used deep learning to identify arrhythmias in the ECG. See [this paper][14] (which used normal 12-lead ECGs to predict which patients would experience AF). There are also several review papers on deep learning and the ECG - [this one][21] looked helpful at first glance, but do have a look around. Update: the toolbox described in [this paper][34] looks like it might be helpful, and [this paper][35] might be helpful for getting a general introduction to the topic.
+- **Identifying AF in ECGs:** See the Computing in Cardiology 2017 Challenge: see [here][15] and [here][16] for an introduction to the challenge, and [here][17] for papers describing code submitted to the challenge, much of which is openly available [here][18]. If you ever wish to use the data from this challenge, then do have a look at the [MATLAB script][26] for importing the data.
+- **ECG quality assessment:** Our work on ECG quality assessment is available [here][22] (and also [here][23], although this paper is narrower in scope). [This][24] is meant to be helpful, although I haven't read it for a while. Whilst [this paper][25] focuses on using multiple signals, rather than just a single-lead ECG, I do like the idea that you can assess signal quality by using multiple beat detector algorithms (see [here](https://peterhcharlton.github.io/info/algorithms/beat-detectors.html#electrocardiogram-ecg-beat-detectors) for example algorithms), and deeming data to be of high quality only if a 'strong' (i.e. highly accurate) and a 'weak' (i.e. less accurate) beat detector agree on where the heart beats are. Finally, [this article][30] describes a technique for identifying transient noise in ECG signals, designed and validated on STROKESTOP data.
+
+### Machine Learning
+
+- **Potential applications of ML in AF screening:** See [this article][32].
+
+## Additional Reading
+
+[This page](https://peterhcharlton.github.io/info/tools/reading/af_screening.html) provides an extended list of reading on screening for atrial fibrillation (AF), much of which may not be relevant to your particular project.
+
+[1]: https://universityofcambridgecloud-my.sharepoint.com/:b:/r/personal/pc657_cam_ac_uk/Documents/SAFER_Engineering_Resources/SAFER_student_projects/Project_Resources/Reading_Materials/Public%20Health%20England%20-%20Atrial%20fibrillation%20prevalence%20estimates%20in%20England%20Application%20of%20recent%20population%20estimates%20of%20AF%20in%20Sweden.pdf?csf=1&web=1&e=Hi3n0e
+[2]: https://doi.org/10.1161/01.str.22.8.983
+[3]: https://doi.org/10.1136/heartjnl-2018-312977
+[4]: https://doi.org/10.1161/STROKEAHA.116.013378
+[5]: https://doi.org/10.7326/0003-4819-146-12-200706190-00007
+[6]: https://doi.org/10.1093/eurheartj/eht280
+[7]: https://doi.org/10.3390/ecsa-7-08195
+[8]: https://doi.org/10.1161/CIRCULATIONAHA.114.014343
+[9]: https://doi.org/10.1093/eurheartj/ehaa612
+[10]: https://doi.org/10.1093/eurheartj/ehz834
+[11]: http://www.mit.edu/~gari/ecgbook/ch5.pdf
+[12]: http://www.mit.edu/~gari/ecgbook/ch6.pdf
+[13]: http://www.jscholaronline.org/articles/JBER/Signal-Processing.pdf
+[14]: https://doi.org/10.1016/s0140-6736(19)31721-0
+[15]: https://physionet.org/content/challenge-2017/1.0.0/
+[16]: http://www.cinc.org/archives/2017/pdf/065-469.pdf
+[17]: https://physionet.org/files/challenge-2017/1.0.0/papers/index.html
+[18]: https://archive.physionet.org/challenge/2017/sources/
+[19]: https://doi.org/10.1038/s41591-018-0268-3
+[20]: https://doi.org/10.1088/1361-6579/aaf34d
+[21]: https://doi.org/10.1016/j.eswax.2020.100033
+[22]: https://doi.org/10.1109/JBHI.2014.2338351
+[23]: https://doi.org/10.3390/ecsa-5-05743
+[24]: https://doi.org/10.1109/TBME.2013.2240452
+[25]: https://doi.org/10.1088/0967-3334/36/8/1665
+[26]: https://peterhcharlton.github.io/info/datasets/cinc2017
+[27]: https://peterhcharlton.github.io/info/algorithms/beat-detectors.html
+[28]: https://doi.org/10.1093/eurheartj/ehaa612
+[29]: https://doi.org/10.3389/fpubh.2017.00258
+[30]: https://doi.org/10.3389/fphys.2021.672875
+[31]: https://doi.org/10.1161/01.CIR.93.5.1043
+[32]: https://doi.org/10.1016/j.cvdhj.2022.04.001
+[33]: http://peterhcharlton.github.io/publication/prioritising_ecgs/
+[34]: https://doi.org/10.1088/1361-6579/ac9451
+[35]: https://doi.org/10.3390/s20040969

safer_projects/safer_data.md

@@ -0,0 +1,77 @@
+---
+layout: default
+title: 3 SAFER Data
+parent: SAFER Projects
+has_children: false
+---
+
+# SAFER Data
+{: .no_toc }
+
+This page provides information on data collected during the SAFER Programme.
+{: .fs-6 .fw-300 }
+
+## Table of contents
+{: .no_toc .text-delta }
+
+1. TOC
+{:toc}
+
+---
+
+## Data Collection
+
+### Recording ECGs
+
+Participants of the SAFER programme are asked to use a handheld ECG device to record 30-second, single-lead ECGs at home. They record approximately 4 ECGs per day for approximately 3 weeks, providing around 84 ECGs per participant.
+
+_The handheld ECG device_
+
+The device being used is the [Zenicor-EKG](https://zenicor.com/zenicor-ekg/) device (see [here](https://zenicor.com/wp-content/uploads/2014/08/MG_1945_low-300x200.jpg) for a picture of the device). This device was previously used in the STROKESTOP studies.
+
+### Automated analysis of ECGs
+
+ECGs are analysed using [Cardiolund's ECG analysis algorithm](https://peterhcharlton.github.io/info/tools/reading/af_screening.html#cardiolunds-ecg-parser-algorithm). This algorithm applies the binary classifications (known as tags) to each ECG signal, which are descirbed [here](https://cardiolund.com/ecg-parser/).
+
+Selected tags are used to identify ECGs which exhibit signs of possible AF, and therefore warrant manual review.
+
+### Participant-level diagnoses
+
+Participants who have at least one ECG exhibiting signs of possible AF are sent for manual review by one or more reviewers (details of this process are provided [here](https://universityofcambridgecloud-my.sharepoint.com/:t:/r/personal/pc657_cam_ac_uk/Documents/SAFER_Engineering_Resources/SAFER_study_methods/Clinical%20review.md?csf=1&web=1&e=qCcgKp), and details of the tags used to identify signs of possible AF are provided [here](https://universityofcambridgecloud-my.sharepoint.com/:t:/r/personal/pc657_cam_ac_uk/Documents/SAFER_Engineering_Resources/SAFER_study_methods/Use%20of%20algorithm%20tags.md?csf=1&web=1&e=NcchiT)).
+
+Each participant is assigned a participant-level diagnosis.
+
+### ECG-level labels
+
+During the process of reviewing participants, cardiologists may label individual ECGs on an ad-hoc basis. Labels include: AF, non-AF, and poor quality.
+
+## Data processing
+
+The SAFER data are extensively processed to produce a clean dataset for analysis. The variables available are listed [here](https://universityofcambridgecloud-my.sharepoint.com/:b:/r/personal/pc657_cam_ac_uk/Documents/SAFER_Engineering_Resources/SAFER_student_projects/Project_Resources/SAFER_variables.pdf?csf=1&web=1&e=0Lq9Vq).
+
+## Accessing the data
+
+Users must sign a _Data Sharing Agreement_ to access the data, which is submitted for review by the SAFER Team. Please ask me for a copy of the _Data Sharing Agreement_.
+
+## Files
+
+### Participant- and recording-level data
+
+The data are provided in the following files:
+
+- **pt_data_anon**: participant-level variables (such as gender and age). Each participant is assigned a unique ID, called _ptID_. The data are provided in two formats, each of which contain the same data: _.csv_ and _.mat_ files.
+- **rec_data_anon**: recording-level variables (such as heart rate). Each recording (_i.e._ each ECG recording) is assigned a unique ID, called _measID_. The data are provided in two formats, each of which contain the same data: _.csv_ and _.mat_ files.
+- **..._diag_key**: The key providing a link between the text diagnoses (such as 'AF') and the numerical diagnoses (such as 1) in either the _pt_data_anon_ or _rec_data_anon_ files.
+
+These files are provided in separate subfolders for each stage of the SAFER Programme.
+
+For an explanation of each variable, see the definitions [here](https://universityofcambridgecloud-my.sharepoint.com/:b:/r/personal/pc657_cam_ac_uk/Documents/SAFER_Engineering_Resources/SAFER_student_projects/Project_Resources/SAFER_variables.pdf).
+
+### ECG recordings
+
+ECG recordings are each provided in their own file, named: _safer[stage]___[measID]_, where:
+
+- _[stage]_ refers to the stage of the SAFER Programme: Feasibility 1 (F1), Feasibility 2 (F2), or Trial (T).
+- _[measID]_ refers to the unique ID for the ECG recording, which is available in _rec_data_anon_.
+
+ECG recordings are provided in WFDB format, which is the standard format used on PhysioNet, and is explained [here](https://archive.physionet.org/faq.shtml#file_types). It is suitable for use with the [WFDB Python Toolbox](https://pypi.org/project/wfdb/). Please see the Introductory notebooks on [OneDrive](https://universityofcambridgecloud-my.sharepoint.com/:f:/r/personal/pc657_cam_ac_uk/Documents/SAFER_Engineering_Resources/SAFER_code) for examples of how to read and process the data.

safer_projects/safer_methods.md

@@ -0,0 +1,38 @@
+---
+layout: default
+title: 4 SAFER Methods
+parent: SAFER Projects
+has_children: false
+---
+
+# SAFER Methods
+{: .no_toc }
+
+This page provides information on the methods used in the SAFER Studies and Trial.
+{: .fs-6 .fw-300 }
+
+## Table of contents
+{: .no_toc .text-delta }
+
+1. TOC
+{:toc}
+
+---
+
+## SAFER Methods
+
+### Reviewing ECGs
+
+Details of how ECGs were reviewed in the SAFER Programme to allocate diagnoses are provided [here](https://universityofcambridgecloud-my.sharepoint.com/:t:/r/personal/pc657_cam_ac_uk/Documents/SAFER_Engineering_Resources/SAFER_study_methods/Clinical%20review.md?csf=1&web=1&e=mcAfBi).
+
+### ISRCTN Numbers
+
+|---|---|
+| Study Phase | ISRCTN number |
+|---|---|
+| Feasibility Study ( _Feas1_ ) | [16939438](https://www.isrctn.com/ISRCTN16939438) |
+| Trial ( _including the Pilot_ ) | [72104369](https://www.isrctn.com/ISRCTN72104369) |
+
+### Use of algorithm tags
+
+The use of algorithm tags is detailed [here](https://universityofcambridgecloud-my.sharepoint.com/:t:/r/personal/pc657_cam_ac_uk/Documents/SAFER_Engineering_Resources/SAFER_study_methods/Use%20of%20algorithm%20tags.md?csf=1&web=1&e=tsnrFN).

safer_projects/safer_projects.md

@@ -0,0 +1,17 @@
+---
+layout: default
+title: SAFER Projects
+has_children: true
+permalink: safer_projects
+nav_exclude: true
+---
+
+# SAFER Project Resources
+{: .no_toc }
+
+This page provides resources and information relating to the [SAFER Programme](https://www.safer.phpc.cam.ac.uk/).
+{: .fs-6 .fw-300 }
+
+---
+
+An overview of the work I have been involved in so far is provided [here](https://peterhcharlton.github.io/project/safer-wearables/), including publications arising from previous student projects (under the 'Publications' heading lower on the page).