Dispatches from the hi-blogs.info krew

MIE2009 – final day September 2, 2009

Today is the last day of MIE2009. I won’t be reporting much, as my flight home is at lunchtime. Last night was the gala dinner, held in a restored beerhall near Sarajevo town centre. A good time was had by all; Izet sang and ended up losing his voice, he told me this morning. I will upload some photos later, if any have come out OK.

The final keynote speaker is Silvia Miksch from Danube University Krems, Austria, talking on “Computer-Based Medical Guidelines and Protocols: Current Trends”. She is a computer scientist, and is addressing the issues from this perspective. She outlined the problem areas from a user perspective, of information overload, transfer of information and knowledge, and assessment of the quality of care. She notes that there are many different definitions of clinical guidelines and protocols, with varying views in different countries. The guidelines tend to be free text, flowcharts or algorithms – they are used with the aim of improving quality of care and consistency of care, as well as cutting down on costs.

My final bit of MIE2009 is Luis Fernandez-Luque’s presentation on “Challenges and Opportunities of Using Recommender Systems for Personalized Health Education” – related to the work he is doing for his PhD. Tailoring health information is an important aspect of medical informatics. Health education has to do with aptitudes and knowledge relating to improving health – can tailor information to the needs of the individual. Traditionally, tailored health information has been in related to specific diseases or health issues – have been 3 parts, ie gathering information on the users, knowledge of the health information available, and then giving the tailored information to the user, often via rule-based systems. Much web-based work in this area, is still based on the traditional methods. One problem now is the plethora of resources available – and it can be difficult to find good quality content.

Now moving into area of information retrieval tools – search engines or recommender systems. Three types of recommender systems – collaborative (based on gathering knowledge for similar users), content (based additionally on items as well as user information) and hybrid. Some examples of health-related systems – HealthyHarlem (community of patients who tag resources); Cancer Sites Recommender (University of Toronto); MyHealthEducator. Such systems do not rely on experts (although some input from them can help), but derive information from the users. Many recommender systems rely just on popularity of items/resources, and may be skewed due to interests of heavy users. Are no ethical guidelines governing development and use of recommender systems, and the sorts of profiling they are doing on users. Recommender systems have potential for use in health – cannot rely just of popularity as a guide, and quality control is needed. Audience question – what is the role of ‘reputation’?