Monday, 27 January 2014

A "Hot or Not" App for Medical Research?

The "Hot or Not" concept is quite controversial and is close to bullying. For those not familiar the user is shown pictures of two people and asked to click on which person they find more attractive. Later versions of this then allow users to comment, as you can imagine these comments can generally be pretty mean.

So how can an application like this be applied to benefit medical research? 

Well I have recently been involved with a project called the James Lind Alliance Neuro-oncology Group. This 12 month project aims to gather questions that people - from patients to carers to medical professionals - have about brain cancer or spine cancer. 

The goal is to gather these questions and then sort them into questions that have been answered and those questions that have not yet been answered - uncertainties. 

Why ask patients and carers, why not leave it to the researchers? Well turn that question on it's head. Why should researchers who have special or vested interests get to drive the research agenda? Why not focus research on what matters most to patients, carers and other non-research folk. 

The Problem

The issue arises in the number of responses that JLA studies generate. There can easily be 1000 + questions asked - in free text form. AKA a bloody nightmare in terms of searching, sorting and filtering. How many variations can you ask a medical questions on a tumor? The number is > 1 the upper limit is unknown, but high. 

The issue is you need an expert to identify whether or not questions are similar. These are medical questions after all. However, two medical professionals could disagree about whether or not a question is the same, and so you need a specialist medical professional to review, and then a peer to review to make sure that there is not vested interests. Then you have to construct a question that covers all the variation so there needs to be collaboration. 

In other words doing this for 1000 + questions is a royal pain in the ass (and very, very time consuming). What's more this is doctors time, which is very expensive.

The Opportunity.

Guess how this process is currently done. Yep you are right, via email and day long face-to-face meetings, which are expensive and annoyingly inefficient. So there is a brilliant opportunity to build something that could help doctors save time and do a better job or categorising and collaboration on these important questions. 

Step in Hot or Not (a variation). 

The Idea

What is needed is a place where the doctors involved in the project can log in to a web-based system and see two questions (rather than people) side by side and say whether they are hot (similar) or not. 

The doctor could then tag each question allowing questions to be group into specialities, i.e. a question relating to brain cancer in paediatrics. Then specialists could view the tagged questions that are relevant for their area. 

Another part of the application would be a straightforward commenting system allowing the doctors to converse and collaborate on various questions.  

Finally, the doctors need to be able to assign a question to a "final version question". 

Some basic metrics and gamification could make this process more enjoyable. I.e. a leaderboard to see which doctor in the group has sifted through the most number of questions or seeing a progress bar to see real time progress. 

The Savings

The number of doctors / health professionals involved in a project is about 10-15, and are usually well dispersed around in the United Kingdom. This makes a day meeting very very expensive, especially since most of the doctors are either surgeons or consultants. 
Cost per doctor (approximately)
Transport - £150
Accommodation - £ 80
Breakfast - £10
Lunch - £20
Venue - £10 
Opportunity Cost (for the day) = £275 (based on a salary of £100,000)

Overall estimated cost for a one day meeting per doctor is £545 or £5450  - £8,175 overall. 

Multiply this by the number of JLA projects per year and it starts to really add up. This cost also excludes the time spent outside of the meeting following up on actions and sorting questions. I would say in total a single JLA project probably costs in the region of £30,000 - £50,000. 

What if this software could make the process of filtering x2 as time/cost efficient as the manual way? Well we could help save £15,000 - £25,000 per project, which is pretty neat. 

The next steps...

- Spec the software 
- Set up GitHub / Project page / Open Source terms. 
- Gather team. 
- Code like mad. 
- Release to test within 1 JLA project. 

If you want to get involved or know someone who might, or want to suggest a cool feature etc then just leave comments below.  

No comments: