|
We have recently released a judgment and decision tool called AnnaLisa. AnnaLisa is the culmination of ten years experience and testing directed towards the development of a generic decision support and communication system grounded in decision analysis – but one that will be accessible - in all senses - to the vast majority of the population. The first system to be developed, CODA, was based in traditional Multi-Attribute Decision Analysis (Pell, Dowie et al 2002) and located in the Shared Decision Making context of clinical medicine. It met the highest prescriptive standards of the technique in respect of both evidential base and utility elicitation, but proved too sophisticated, too focussed on specifically health-related outcomes (as measured by QALYs), and too demanding for the current organisation and resources of the clinical situation. It was also highly decision-specific, being designed solely to assist the prophylactic oophorectomy decision of a woman who had already decided to undergo hysterectomy. The second, MyWay (French, Dowie 2007), shifted the modelling base to Multi-Criteria Decision Analysis because of the greater flexibility in outcome specification and attribute customisation this offers to all parties and stakeholders in a decision context. It proved a clear success in this key respect, as well as others, during its piloting in the context of contraceptive guidance to teenagers. However the program was dependent on bought in software which would have made its further and wider application problematic and expensive and was still in some ways over-engineered. Annalisa was developed in the light of these experiences and considerations, with the objective of offering maximum flexibility and simplicity compatible with the demands of MCDA - but doing so at levels of price, analytic complexity and user requirements that were much more appropriate to the general population than the available softwares using this theoretical basis, including notably the Analytic Hierarchy Process as implemented in Expert Choice. There is nothing new in Annalisa conceptually but the implementation offers the user an interface with features that we believe make it very attractive to anyone seeking to use MCDA in teaching, research or practice. In many ways Annalisa is properly described as MCDA-lite, since this designation captures the primary intention of reaching the many deciles of the population that others have manifestly failed to do. In many cases this is because the developers have not seen this audience as their ‘market’, in others because they have seriously over-estimated the user requirements that are compatible with widespread public use. But referring to Annalisa in this way should not be interpreted to mean that it cannot also be used where rigour as well as practicality is essential, wherever a linear compensatory expectational model is prescriptively defensible – which it is in the vast majority of cases. Annalisa permits many different trade-offs between these two high-level attributes and, moreover, can help clarify that trade-off in the course of addressing the task of deciding how to decide. Annalisa has therefore been designed specifically to function as a semi-analytical decision support and communication tool that is intuitively accessible to the ‘ordinary person in the street’ - and to the ordinary professional or official in an agency type role. As can be seen from the diagram of the default screen below, the three panels (from the bottom up) represent the evidence base ( Ratings of up to 10 user-provided options on up to10 user-provided attributes), the value base (user-provided weightings of the attributes) and the option ratings (score produced by applying weightings to ratings and summing). All ratings may be entered either numerically or graphically and either in percentage or decimal form. All three panels may be made instantly visible or hidden as desired. (It will often be desirable to hide the scores panel while eliciting the weightings of the attributes and there is no need for the patient to see the ratings panel unless they seek the information it contains.) The underlying scoring principle is of maximising expectation, though other parameters can be incorporated through modification of the criteria. It is envisaged that Annalisa will typically be used with default ratings (i.e. evidence base) for defined patient/client groups, though when used by the professional in the consultation situation he or she always retains the ability to tailor ratings to the individual they are with where this is felt to be desirable. The file can be saved in txt, csv and jpg formats both for record purposes and in order that communities of practice may develop and disseminate Annalisa (.alt) files for discussion and evaluation purposes.
Ian Pell, Jack Dowie, Aileen Clarke, Andrew Kennedy, Vanita Bhavnani (2002). Development and preliminary evaluation of a Clinical Guidance Programme for the decision about prophylactic oophorectomy for women undergoing a hysterectomy Quality and Safety in Health Care 11 32-39
Rebecca French, Jack Dowie (2007) Using decision analysis to help young people with contraceptive choices. In Royal College of Obstetrics and Gynaecology Teenage Pregnancy and Reproductive Health (London: RCOG Press, forthcoming)
|
THE BASICS 
