# Day 1 : Dissertation project requirements

## Resources for literature review

**Tools guide:**
https://libguides.cam.ac.uk/Official-Publications/referencing

**Referencing style guide:**
https://www.cambridge.org/core/services/aop-file-manager/file/610be5e3a1acae0932de3cca/Cambridge-Reference-Styles.pdf

---
## Resources for Health Data Research Project

**R and Rstudio**
https://rstudio-education.github.io/hopr/starting.html

**RMarkdwon**
https://rmarkdown.rstudio.com/

**Python and Jupyter notebooks**
https://docs.jupyter.org/en/latest/install/notebook-classic.html

**Visual Studio Code**
https://code.visualstudio.com/download

**Github**
https://docs.github.com/en/get-started/quickstart

**Gitkraken**
https://www.gitkraken.com/download

---
## Example study
### Risk of thrombocytopenic, haemorrhagic and thromboembolic disorders following COVID-19 vaccination and positive test: a self-controlled case series analysis in Wales

https://www.nature.com/articles/s41598-022-20118-6#Abs1
link to the research pipeline and code of this study:
https://github.com/HDRUK/DaCVaP/tree/main/Wales

---
## Reflection point 1
Consider your research proposals and try to reflect on the following points

**Where is the gap**
Do I have sufficient evidence to defend my proposed idea?

**Is it feasible**
Do I have the variables required to analyse my outcome of interest?

**Method**
What is the best approach to address my research question? statistical evidence

---

## Reflection point 2
Does your research proposal cover clearly the outcome, exposure and method?

**Outcome**
What is the clinical endpoint I am assessing?

**Exposure**
Which factors might influence the endpoint?

**Method**
Statistical Analysis Plan

---

## Providing summary data

Useful packages and tools:

### for R users

**Tableone**: https://cran.r-project.org/web/packages/tableone/vignettes/introduction.html

**gtsummary**: https://www.danieldsjoberg.com/gtsummary/

### for Python users

**Tableone Python:** https://pypi.org/project/tableone/

*Project example* https://academic.oup.com/ehjopen/article/2/6/oeac066/6835611?login=false

---
# Day 2: Dissertation proposal presentations

| Day-Time | Presentor |
|---------------------|-----------------------|
|Thur(9:40-9:55) |Adedayo Adeyomoye |
|Fri (10:15-10:30) |Nana Atefah |
|Fri (10:30-10:45) |Mary Stuart |
|Fri (10:45-11:00) |Antonie Deysel |
|break |15min |
|Fri (11:15-11:30) |Ruhui Wang |
|Fri (11:30-11:45) |Joseph Olivelle |
|11:45-12:00) |Samuel Gunning |
|Fri (12:00-12:15) |Brianda Ripoll |presented first
|Group discussion |Until 13:00 |
|Lunch |back at 13:45 |
|Fri(13:45-14:00) |Venkata Chada |
|Fri(14:00-14:15) |Satpal Shekhawat |
|Fri(14:15-14:30) |Muhammad Ambia |
|break |30min |
|Discussions |untill 16:30 |

---
# Presentation Feedback Notes

Please add your initials followed by your notes/comments for consideration under each person's name

---
## Adedayo Adeyomoye
### Title: What is the treatment success rate using the national Tuberculosis data for the completely treated and cured TB patients in Nigeria amidst different possible treatment outcomes?

Peer-feedback notes availabe via the zoom chat history below:
* 09:41:45 From Healthcare Data to Everyone:
let's get started with your presentation
* 09:41:53 From Healthcare Data to Everyone:
we can hear you very clearly
* 09:54:01 From Healthcare Data to Everyone:
Opening it to the room for peer-feedback
* 09:54:49 From Healthcare Data to Everyone:
What sort of data you are hoping to get from the data provider?
* 09:55:17 From Healthcare Data to Everyone:
Q from TM: What sort of data you are hoping to get from the data provider?
* 09:57:47 From Healthcare Data to Everyone:
Q from Satvel: what is your measure for success rate? how are you defining the success
* 09:59:56 From Healthcare Data to Everyone:
follow up Question: are you sure the measurements and outcomes are included in the dataset?
* 10:01:45 From Healthcare Data to Everyone:
and last one on feasibility: the success rates are defined similarly for all subjects in the dataset? do you need to do any further manipulation to harmonise and achieve the final measures for the project?
* 10:03:41 From Healthcare Data to Everyone:
Note for consideration: treatment adherence is a highly challenging factor to assess via Electronic Health Records
* 10:04:54 From Healthcare Data to Everyone:
Does Nigeria have Direct Observed Therapy?
* 10:05:36 From Healthcare Data to Everyone:
Info about prescription and dispensing of drugs, maybe specifically TB treatements
* 10:08:33 From Healthcare Data to Everyone:
Great yes
* 10:09:12 From Healthcare Data to Everyone:
One more please: visualisation is this going to be a TB dashboard or static figures in R - who is the audience?

---
## Brianda Ripoll
### Is it the time for the Cardiothoracic Surgery Training in the United Kingdom to adopt a regular simulation training programme for cardiac surgeons? 

Peer feedback notes:

* summary: Following 13 candidates for assessing surginal skills: Nominal + fundamental principles + Delphi method
* You mentioned these findings are applied to UK CT training, how generalisable are these results to those who are training in a different health system?
* FT- Hove patient outcomes been assessed for this cohort?
* FT- Comparison population? was these trainees been matched and compared to a matched cohort of trainees who have not been on the programme to compare the performance between the two?
* Think about control groups and how to acknowledge the strengths and weaknesses of the conclusions you are drawing and how these could be built on in clinical practice in the future.
* SSS- presentation illustrates the effort and work gone behind this. control groups will help evidence on proving improvement is attributable to simulation. how will you use your data to influence deaneries to change to non differential data?

---

## Nana Atefah

### Is it time to consider statutory regulation for management in NHS in a bid to curb the menace of patient injury as a result of managerial negligence.

Peer feedback notes:

* FT- Adding the system specification will make the case clearer
* FT- Provide where your proposal fits within the healthcare vision, what is actually in the plans of UK governance
* FT- What would be involved in impelementation of a set of new regulations? what does this dissertation work contributes to the requirements?
* FT- Providing and working on a case study of how introducing this or partial bits of this proposal would bring a change?
* What are the barriers and preservations for operationalising these models
* SSS- Topic is excellent, relatable and looking at accountability for actions for managerial roles. Focus on describing the process and how will you provide evidence on why it is needed? Is work place appraisals not fit for puprose and can that not be developed to meet this need?
* Are there other health systems with this type of regulation that can provide a comparison in terms of implimentation and effectiveness?
* Are there current case studies where manager-led decisions have been better for the patient as a flip side to this argument?
* Great topic. Much like any regulation it would be interesting to know what the disadvantages of introducing regulation of managers will be (e.g. regulating managers might freeze a lot of the functionality of the system)? (S.G)
* HW: comparative study if regulatory and complaince actions/requirements across industries, such as medtech, pharma. and also consider other areas that could be somehow relevant, such as ethics and responsibility of AI, its such a new area and the way they define ethics and reaponaibilities that impacts human being could be inspirational

---
## Mary Stuart
### Insert the Title

Peer feedback notes:

* SSS- in UK we do look at Arthritis, sleep apnoea, hypertension,CVD and Diabetes as co-morbidities along with what other weigh loss methods have been tried when applying for Independent funding requests for Bariatric surgery. BMI levels are also considered.YHEC (York Health Economics Consortium) does cost impact analysis and will be a good resource for your literature review. CCG are no longer there, they are now ICB's.
* SG - for inspiration on which conditions to focus on you could look at metabolic syndrome, the constellation of symptoms/signs/investigation findings and the most common resultant comorbidities (e.g. coronary artery disease, stroke).
* FT - for lit-review componant: great overview of all existing and known corellations between weight and comorbidities but would suggest a focus on a specific disease area i.e. Heart Failure patients, Women of Child Baring , Cancer patients as each one have many other factors involved affecting possible care pathways
* FT - adding to above cost-effectiveness analysis for a certain patient group. Health economics analysis rather than lit-review, business case to influence CCG or ICB(Integrated Care Board)
* TM - if you start with a focus on one co-morbidity it is easier to expand your coverage if the literature review or the data side look viable, rather than trying to narrow down from a wide starting point.
* BC - can consider the mediating effect of patients on GLP-1 agonists on healthcare costs
---
## Antonie Deysel
### Barriers to Dental Treatemnet for Patients with Mental Health Problems and Anxiety: Strategies for Improvement

Peer feedback notes:

* FT - Nice formulation of solutions - would suggest focus on patient outcomes for specific age group i.e. pediatric dentistry
* FT - Outlining the specific needs of this subgroup of people with specific mental health problems
* SSS- great context building and have explored wide areas to explore and look at solutions. you may need to focus on ome or few to manage word limit. building on evidence base on which solution has most impact as will help other dental practitioners.
* TM - there's potentially a lot of different factors to cover here. It may depend on what is published but a narrower focus e.g. on one specific intervention might be more manageable.

---
## Ruhui Wang
### HCM (in development) - mose likely to be searching around how well HCM is documented in EHR, and what are the predictive symptons that could be used as a diagnose supporting factor

Peer feedback notes:

* BC - How well do you expect the symptoms of HOCM to be tagged within the EPR? May want to look into the EKO Duo device (used for detecting arrhythmias, HF-rEF, and structural heart disease) (https://www.imperial.ac.uk/news/249316/ai-stethoscope-rolled-100-gp-clinics/). Is there clearly defined treatment pathways for HOCM patients? Availability of echo to confirm diagnosis? May need to implemented in order to see meaningful change in practice.
* TM - your original idea was focused on positive change i.e. designing a predictive algorithm. Is there a way to utilise the EHR data to still focus on positive change as this will enhance the "so what" element?
* FT- Question please - what do you mean by statistical relevance of HCM relater symptoms?
* FT - Patient journey analysis: this require a robust definition of "A Cohort". Then for this cohort what would be considered as a formal diagnosis in secondary care data? eco confirmed Cardio myopathy? from discussions I would suggest:
* Cohort of confirmed HCM patients
* Investigating pre- and post- diagnosis pathways including surgical intervention, preventative medications, etc.
* FT - developoing a risk prediction scheme requires evaluation of some contributing factors prior to formal diagnosis of HCM
* FT - lit review to focus on what factors are involved in pre-diagnosis assessment of HCM: the aim would be to come up with a list of clinical conditions that would propose a risk scheme
* SG - could look at biscupid aortic valve as an alternative but think you could still build predictive model based on symptoms and demographics (as demographics different for HCM and HTN) and subgroup analysis producing regression model with sensitivity and specificity for identifying HCM on ECHO based on age.
* SSS- as Verbally discussed, looking at examples of PIN point test, QB test and apply principles om predcitive algorithm. will all depend on data availability and limitations

---
## Joseph Olivelle
### Has specialism in Podiatry led to increased medical negligence claims?

Peer feedback notes:

* SSS-Context is useful, complaints and claims are multifactorial so focussing on one aspect will help Lit Review and any measurements you want to do.
* FT - would be keen to see a development of the idea to allow derivation of a proposed solution that then can be critically evaluated within the body of dissertation
* TM - worth considering if there is a comparative specialism that you could look at to see if similar/different trends occur in these as in podiatry.

---
## Samuel Gunning
### Predicting treatment response in the heterogenous heart failure with preserved ejection fraction subgroup using synthetic retrospective single-centre data.


Peer feedback notes:

* Focussing on specific treatments? e.g. SGLT2-inhibitors? Is this acute heart failure necessitating inpatient admission? If focussing on diuretics, more looking at immediate outcomes e.g. LOS/30-d mortality vs. long term prognosis? All should be clearly available data?
* SSS - clearly thought process, how will you measure treatment reaponse? what measure or scales will be used. There is ongoing research on exercise programmes to this cohort so would you include that in treatment options.
* FT - for validationg of the method used by Woolley et al. the population needs to be matched i.e. CHF - are we comfortable to use HFpEF instead, in which case we can not call it a validation study but we can certainly apply the method they've used.
* FT - are we able to derive the 4 clusters with likely less data on 363 biomarkers?

---
## Venkata Chada
### Assesing the suitability of the ISARIC 4C score to predict mortality in older adults admitted with COVID-19 using synthetic inpatient hospital data

The 4C Mortality Score and 4C Deterioration models are risk stratification tools that predict in-hospital mortality or in-hospital clinical deterioration (defined as any requirement of ventilatory support or critical care, or death) for hospitalised COVID-19 patients, produced by the ISARIC4C consortium

Peer feedback notes:

* SSS- clearly thought our, have considered limitations of data and what is achievable. will you look at data or evidence on functionality of elderly to provide arguement against just using 4c as a predictor
* FT - the ISARIC 4C score is developed for in-hospital mortality which should fit nicely within the scope of EPIC
* FT - suggested result: https://thorax.bmj.com/content/thoraxjnl/77/6/606.full.pdf
* FT - account and stratify based on availability of COVID-19 vaccinations role out.
* HW - patient cohort selection: some disease such as cancer have different scoring system, worth thinking through what cohort forms your inclusion criteria (then either exclude, or perform mapping of scoring systems across other disease definitions)
---
## Satpal Shekhawat
### Is Digital Consenting in primary care a time efficient process?


Peer feedback notes:

* FT- What sort of information is provided to patient as part of the consenting process?
* FT- do we have any patient feedback? on ease of use and digital litracy
* FT- A great deal of development went in structural design of the digital consenting until it reached the practical operation level: I would hihgly recommend covering all those in relevant sections in the dissertaion. Three recommended themes would be: 1)Care provider 2)Patient and public 3)Practice level delivery
* Suggesting a change of title showcasing the amount of work went in this rather than proposing a question
* TM - is there any patient interaction/data - is ethical approval needed?
* TM - this is a link to a scoping review on electronic consent in clinical practice - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10335420/

---
## Muhammad Ambia
### How integrated data can support better insights in asthma prevalence for children and young adults

Peer feedback notes:

* FT - The litreview is setting the scene very nicely with some stats, would suggest adding some details on what would be the contribution of integrated data? (it looked like these already exist) is there specific type of integration i.e. health and schools data?
* FT - there are healthcare data sources that hold asthma as a recorded comorbidity, would recommend exploring also a key word search for Asthma in: https://web.www.healthdatagateway.org/search?search=asthma&datasetSort=latest&tab=Datasets
* FT - in litreature review then focus on the existing resources for Asthma and is there a clear contribution in better estimating prevelance of disease if we get all of these linked together
* There is a lot of potential information you can cover in the literature review. You might want to consider how to focus on some of these more specifically so that you can give a critical interpretation.
* BC - Do you think you can find a link between allergen data and asthma data? I imagine most of the allergen data will be about drug allergies (e.g. penicillin), which is unlikely to be linked to asthma. Other drugs (e.g. beta-blockers & NSAIDs) may have an association with asthma, but probably only in a handful of cases (if any)

---

# Help us to improve
## Feedback on the day
### Positive notes about the day
* Plenty of time given for presentation and feedback from peers. it was very useful to shape with the work i need to do for the dissertation.
* Really benefitted from peer input into the dissertation development
* Glad for the opportunity to run through our presentations and receive feedback, helped to gather my thoughts and make a plan for how to proceed
* Peer feedback and individual supervisor both very helpful. Good to plan further steps as well including review of Epic with Ronan and further supervisor sessions and contact with the group. Expectations clearly outlined for April.
* Great real time feedback from teaching team - added a lot value to the works
* Really good feedback and overview of the dissertation structure!

### How did you find proposed content and structure?
* All content very relevant to dissertation writing and formulation of ideas. Structure of the two days was well formulated to get most yield fromromrom
* content was relevant and helpful with the programme. there was overlap between healthcare and clinical medicine group and there are some differences between the contents for both cohorts
*The structure of the day was really good with a lot of time allocated to peer feedback which was really helpful
* Overall, good structure. At times it felt like the groups could have been separated so that information was more specific (without being redundant to one or other group)
* Found it helpful to go through the proposed outline for the dissertation. Would have liked to have spent more time on structure and content of dissertation (e.g. indicative word count for each section)
* Very helpful. Placed things in context.
* good to get peers to review your proposals as they have good insight
* I wish the course team would have shared some feedback on the proposal that we submitted - arguably we could have done addtional researches when it comes to changing research directions and data requirement

### What could be improved your contribution for FT
* using a session to walk through a few examples of successful submissions will help visualisation of what is required.
* agree with the above comments
* Nothing.
* Nothing - it's great to hear you adding a lot commentaries during presentations - super useful - particularly when you add value on topics that we haven't thought of based on your knowledge in academic writing and in healthcare data and analytics
* Nothing - great to have your input on our projects.
* FT has been really helpful. The overview of the referencing process and scoring of the dissertation has given a great starting point. Due to time constraints some of the demo's on using github and the statistical pipeline couldn't be covered. Would be great if these could be added to VLE or maybe having a seperate online session?
* Connection to the wifi could be better please
*
Shorten the time from the end of this MSt into progression to a PhD programme.

Ensure this MSt is not the end of the academic road to this cohort. Enrure progression into PhD please.

### What could be improved your contribution for TM
* the work on marking schedule for clinical medicine group was very useful. examples of submissions will be useful
*
*
*
* Helpul contributions and suggestions for how to refine and proceed with dissertation
*
*
* helpful contribution during presentation - added value on potential scope refinement and/or extention
*
* helpful comments
*
*

### What could be improved your contribution for ROL
*
* Little more clarity on the types of data available to us, how it can be accessed etc
* Earlier data insight to help avoid wasted work on topics
* Would be great to showcase what Epic data contains
*
*

---
We are very grateful for your contribution
Fatemeh, Tom, Ronan on behalf of the MSt Healthcare Data and Clinical Medicine teams

:smile: likewise!

---
# Contribute to the course
Please log your interest if you are interested to join our alumni cohort of contributors. We will have a range of opportunities and your contribution is highly valuable to us as those who had first hand experience on the course.

## contact details for keeping in touch

* Name | your prefered mode of contact | Happy to be contacted | Happy to contribu emailte to course |
* Fatemeh Torabi | email: [e-mail] | :+1: (Monday & Thursdays) | Yes :+1:
* Mary Stuart | email: [e-mail]; mobile: [phone] | Yes | Yes |
* Bharadwaj Chada | email: [e-mail] |Yes|Yes
* Antoine Deysel | email: [e-mail] |Mob:[phone]| yes| yes
* Hugh Wang | [e-mail] 07853129707 | Yes | Yes
* Nana Atefah | email: [e-mail]|Yes to be contacted|Yes to help with course.
* Satpal Shekhawat |-email- [e-mail]|mobile- 07849605933| Yes- to be contacted| Yes- to help with the course
* joseph olivelle email [e-mail]
07881520677
* Brianda Ripoll - [e-mail] 07474956555
* Samuel Gunning - [e-mail] - 07712336527 yes yes and yes and yes.
* Muhammad Ambia - [e-mail] / [e-mail] - 07503174416 - Yes to all

---

Tuesday and Wednesday : 23rd & 24th
------------------------------------------------------

# Common variables from EPIC data for MSt dissertation projects

useful link from BC: https://www.youtube.com/watch?v=xGYFDrORpzA

**Baz Chada**

* Gender
* Postcode
* Date of death (primary outcome)
* Ethnicity
* Admission ID
* Date of admission (to calculate LOS - secondary outcome)
* Date of discharge (to calculate LOS - secondary outcome)
* Admission method
* Admission source
* Discharge destination (may show "n/a - patient died")
* Date of transfer to ITU (secondary outcome)
* Date of discharge from ITU
* Problem Date noted (to confirm Covid Dx)
* Problem/ Diagnosis SNOMED code (to confirm Covid Dx)
* Med Hx - Problem/ Diagnosis name (for comorbidities - can be specified)
* Med Hx - Problem/ Diagnosis SNOMED code (for comorbidities - can be specified)
* Problem/ Diagnosis name
* Drug name (ideally to confirm proof of vaccination)
* Respiratory rate
* Oxygen saturation
* Heart rate
* Blood pressure systolic
* Blood pressure diastolic
* Height
* Weight
* GCS (or AVPU as proxy if not available)
* Test name (Components of ISARIC-4C score are CRP & Urea. Other useful prognostic indices may be Hb, WCC, Plts, lymphocytes.)
* Test result
* Test units
* Result date

---
**Samuel Gunning**

* Problem/ Diagnosis SNOMED code
* (Diagnosis of Heart Failure with Preserved Ejection Fraction)
* Problem/ Diagnosis name
* Gender
* Age
* Ethnicity
* Admission ID
* Discharge destination
* Date of admission (to calculate LOS - primary outcome)
* Date of discharge (to calculate LOS - primary outcome)
* Admission method
* Admission source
* Admission specialty
* Number of hospital presentations over past x years
* Weight (to calculate daily weight loss - secondary outcome)
* Problem/ Diagnosis SNOMED code (other comorbidities - I can specify probably)
* Problem/ Diagnosis name
* Fluid balance - fluid intake and output (as a proxy for diuretic effect - secondary outcome)
* Daily observations - primarily oxygen saturations/respiration rate (as a proxy for diuretic effect - secondary outcome)
* Blood results - primarily daily renal function (test name, value, units and date)
* Diagnosis of AKI (during the admission - secondary outcome)
* Medication names, doses, dates started and stopped
* Date of death
* ITU admission date

---
**Mary Stuart**

* Patient ID (assume this is made up!)
* Gender
* Date of death
* Ethnicity
* Admission ID
* Date of admission
* Date of discharge
* Admission ID
* Admission method
* Admission source
* Discharge destination
* Problem/ Diagnosis date noted
* Problem/ Diagnosis date resolved
* Problem/ Diagnosis SNOMED code (primary outcome obesity, secondary outcomes T2D, hypertension, possibly others)
* Problem/ Diagnosis name (primary outcome obesity, secondary outcomes T2D, hypertension, possibly others)
* Medical History SNOMED code (primary outcome obesity, secondary outcomes T2D, hypertension, possibly others)
* Medical History name (primary outcome obesity, secondary outcomes T2D, hypertension, possibly others)
* Medical History date of diagnosis
* OPCS name (looking for bariatric surgery)
* OPCS code
* OPCS date of procedure
* Surgical History name (looking for bariatric surgery)
* Surgical History code
* Surgical History date of procedure
* Obs Height
* Obs Weight
* Obs Date of each

---
**Muhammad Ambia**
* Data on Asthma admissions/attendances in hospital (inpatient, outpatient and emergency department/ITU) - including Admission/Discharge dates/Admission Method/Discharge destination etc
* Demographic data – Age, Gender, Ethnicity, LSOA
* Medication data – Asthma medication use
* Medical History/Family History of Asthma
* Other OPCS/ICD-10/SNOMED Codes – to check co-morbidities with regards to the asthma admissions
* Observation data – personal factors like BMI, respiratory rate, oxygen saturation that can trigger asthma
* Lab results related to asthma biomarkers (Bronchoscopic samples, Induced sputum, Blood, Urine, Exhaled gases)
* Allergy data – Review relationship between asthma admissions and allergens
* Any additional data related to Asthma on EPR. For example, asthma staff training or feedback scores from patients/ family on Asthma care provided

---
**Hugh Wang**
- Basic patient info: patient ID, domographic data (age, gender, ethinicity)
- Medical history / past diagnosis: looking for past records of congenital cardiac disease/HCM diagnosis and/or symptons
- Admission (dates, method, source, specialty, discharge results)
- Problem/ Diagnosis (code) of the admission
- Pathological symptoms data such as chest pain, arrhythmias, short of breath, palpitations, fatigue & light headedness, dizzy spells & fainting, heart murmur, swelling in the feet, ankles, legs, belly or neck (the data may sit in other diagnosis that linked to specific SNOMED codes)
- Test results (ECG/Echo outcomes)
- Referal data (if possible): looking to understand if patient is been referred to a cardiac specialist for further testing (such as genetic testing)
- (optional)Medications given during admission - looking to pontentially identify cardiac disease related medications

End of document