Second Workshop

In the second blog for the second workshop of Tutorials in Health Data science I am going to answer these questions:


  • What did I know already about the subject that was discussed in the workshop?
    • I was and I am very familiar with the dashboards as I am a software developer and I have worked for a few years on different web based applications in different industries. Querying large database and display a set of useful data specifically for different types of users is very common.
    • Also, on the section of various types of knowledge and the conversion of one type to the other, as I have done a M.Sc. in Knowledge and Information Systems Management, I am very familiar with the concepts. I have a few opinions on the way those types have been raised. But mostly, the presentation matched with my previous knowledge. For example, Tacit knowledge was mentioned but some of the examples can be categorised as an implicit knowledge.
    • Most of the information which was given on the effectiveness of CDS was discussed on the first workshop.
  • What are the most important things that I have learned?
    • On the section for knowledge sources for CDS, there was a diagram of CDS architecture and how knowledge engineering would feed into the system. That diagram was one of the most important things which I learnt in the workshop.
    • Also, inference processes for CDS had some important concepts. Although, most topics were mentioned in the given book chapter which was read a couple of weeks before, still the section machine learning and new approaches were concepts that was a mind refreshing for me.
  • What are the things that I don’t (completely) understand?
    • I have raised a question with the Professor during the workshop and I followed it up with an email later on. The way ontology described during the lecture, gave this impression that it is a knowledge representation which is created by human being. However, in the recent researches, some trends are into the creation of onologies using the deep machine learning. This means, machine learning can represent its knowledge using an ontology which can be fed to another system as a knowledge base.
  • Has the content raised any new questions for me?
    • On the section which we talked about the effectiveness of CDS, I was eager to hear more about other types of evaluations such as accuracy, safety and usability.

IT systems in Healthcare and some questions

Computerised Decision Support or Clinical Decision-Support Systems (which is still not clear to me which one we normally refer to, but it is not majorly important) are systems which provide right knowledge and information in the right time to the right person who might be a clinician, patient or any other stakeholder. On the other hand, EHR systems or Electronic Health Records systems are some repositories which hold all the health related data for an specific patient. It is clear that CDS systems can be more powerful if they have access to EHR.

However, there are different levels of support and already there have been some attempts to develop some systems to help the process of decision making in healthcare, I was a bit shocked when I obtained more insight in this area. Since my background is in computer science and I have been in programming industry for some years, I was expecting to see more of recent technologies in this sector. I was shocked when I heard even in the United States of America (home to the Silicon Valley which is pushing the IT boundaries), paper is still the most trusted in some health organisations. I became more aware of the impediments and limits such as systems are not easily connected to each other. Or in another case it has been mentioned that users are not very happy to use these systems because change is always difficult.

While I was reading the two suggested chapters and during the workshop session some ideas came to my mind which led me to investigate more and check if anyone has got an answer to my questions which I share it here:

– At this point of time, cloud computing and cloud storage are very popular. Amazon,  Microsoft and Google have services (AWS, Azure and GoogleCloud) which are hosting many important and famous websites such as LinkedIn, Spotify, Ebay, etc.
By using these platform to host healthcare services, owners can spend the effort on the implementation of systems and they can ignore any infrastructure concerns such as availability or consistency. Is there any EHR or CDS system which is currently serves from a cloud? Is there any major challenges to achieve this?

Based on my little investigation, there is a new trend of having cloud-based systems in healthcare. According to ‘Healthcare IT News’, more and more healthcare providers are using cloud as ‘flexible platforms help them to prepare for population health management and value-based care’.

– My other concern was about using APIs in healthcare. Communication between the systems is vital as it various systems hold different information. It has been mention few times during the session and within the chapters that there are obstacles for these connections but it was not clear to me. Again, use of APIs within the computer programming world is very common and there is no need for standardised connections as each two systems can be easily connected. Is this being used?

Another brief investigation showed me that although it has been suggested by US’s Health IT government to use APIs, many EHR system still do not accommodate that.

– Another important topic which I thought about after the workshop is the use of UX. Nowadays, big IT corporations invest a lot on the UX research and design in order to achieve a user-friendly system. On the other hand, more and more people are using computers and internet which means they are getting used to the designs and patterns. Based on World Internet Stats, more than 51% of the world’s population now have internet access. This means although we need to still be careful (as probably those who still do not use internet and computer are either old or from poorer countries which in both cases they need more healthcare) there is an opportunity here since the system education cost will be eventually moved to the design stage (instead of having a separate step for educating users in order to know how to use the system).

Beside all these points, in the Chapter 12 of pre-sessional readings, under the section about ‘Costs and Benefits’ there was a mention of open source systems. Although many of open source systems are accessible for free, being open source does not necessarily mean that it is free. I am not sure why this has been mentioned in that section.