Data & Knowledge-Driven Decision-Making

Guest post from Juha Pajula from VTT

How often are decisions made with gut feeling and why? Various digital records are collected from most of the population in modern digital society: why are those records not used to help make more aware decisions in policy making? In a common sense, it should be fairly simple to check how many people have used the health services in certain period of time, or which age groups of the population are the most passive in some particular region. Probably one reason for the failure to use the collected information is that typically only limited tools and options are available to get the information, and even less options are available to utilize the data from multiple sources without being statistician or data scientist. Only a few policymakers or their support personnel have this kind of expertise.

In MIDAS EU project the MIDAS Platform will include three main components, which will also support some additional analytics systems outside the main system. With the main system the first component delivers the data integration layer, which serves secured data from multiple sources. The second component is the analytics system which includes the needed analytics; and the third component is the MIDAS Dashboard. The MIDAS Dashboard is the only component visible to the end user, where the key aspect in MIDAS Platform is that the data stays where it is stored and the end user needs only to define what kind of results he wants. With the MIDAS Dashboard users do not need to play with tabular data or fight with analytics tools – they just select the preferred predefined analytics for the variables they are interested in and the preferred visualisation for the results. The smart UI of MIDAS Dashboard will help the user to make sensible selections, and actually the UI allows user to select only analytics and visualisations which are sensible for the rest of the selections. From a security point of view, the end user never gets the raw data, or actually any data, the UI shows only results from the analytics and all data stays safe and untouched behind the data virtualisation and analytics layers.

How does the MIDAS Dashboard then work? The MIDAS Dashboard UI is actually a dashboard generator with a smart wizard to generate various analytics widgets. Each widget defines independently data selection, analytics and visualisation. The user of the dashboard can freely create, modify and remove widgets from the dashboard to bring visual results from various analytics into a single dashboard. This enables user to investigate multiple sources of information at once and visualise them in single place. All analytics in MIDAS Dashboard are pre-defined in the analytics component, so the user does not need to know how to define them or how to visualise them. Mainly the user should have only an idea that what kind of analytics is needed with which data, and how the results should be visualised. As health data are typically extremely sensitive, the system is designed to include a safety function, which refuses to show results if they would point to or are sourced from too small a cohort of people. This should not cause problems for end users, as policy-making should not be targeted to small cohorts but large populations.

Policymakers may have only a short time to prepare their proposals and often they rely on other personnel who actually find out the necessary knowledge to support the decisions in hand. For this reason, MIDAS UI will support different user types. In practice, policy makers can have their assistants and other support personnel, who understand the role of data and how to analyze it. These supporters can use MIDAS Dashboard to create the decision support views for different questions, and then share them to the policymakers. In MIDAS Dashboard, there will also be options to use external Twitter, global news and MEDLINE analytics, which utilize global data. These enable possibilities to study the public opinion and scientific basis for the policies which are under consideration. Together with health policy related data, Twitter campaigns, news analytics and MEDLINE abstract search, the health policy decisions can be supported with actual data and hard knowledge!

Reference: MIDAS deliverable D5.1 Visual analytics tool(s) concept V1