In many medical settings precise numerical data for all findings of interest either is unavailable or is too timeconsuming to measure. In such situations where a mixture of exact and approximate data is available, a model which can reason with such data and produce output of precision proportional to the the precision of the available findings is desirable. We are developing an acute cardiovascular model which can reason with mixed qualitative and quantitative data, the principles of which can be applied to any mixed-data situation. Our objective is to construct a computationally efficient, first-principles model of the cardiovascular system's response to blood loss and fluid replacement. Such a model can be used to estimate the volume of blood loss within therapeutically acceptable ranges and predict the future clinically-relevant cardiovascular states of the patient, i.e. to predict initial blood transfusion requirements within -t-1 unit (500cc), and to predict when the patient may go beyond Class II shock (1500cc blood loss or 30% blood loss) either in the absence of remedial procedures or in the presence of fluid replacement or other therapeutic actions.
The 28th International Workshop on Qualitative Reasoning (QR-15) presented advances toward reasoning tractably with massive qualitative and quantitative models, automatically learning and reasoning about continuous processes, and representing knowledge about space, causation, and uncertainty. The technical track included two invited talks, 11 oral presentations, and 5 poster presentations.
This abstract describes an approach to the organization of qualitative design knowledge about mechanism's functioning. This knowledge is being formed on the basis of analytical results obtained in quantitative simulation. Smooth response functions generated by simulation system are approximated by differences of first three orders. Attention is drawn on capabilities of sign combinations of the differences to provide qualitative reasoning when searching for synthesis solutions. The tasks arising are examined from this viewpoint: extraction of qualitative features of function, their classification, setting dependencies between parameters and features, derivation of tendencies of change of the features.
When given a task, an autonomous agent must plan a series of actions to perform in order to complete the goal. In robotics, planners face additional challenges as the domain is typically large (even infinite) continuous, noisy, and non- deterministic. Typically stochastic planning has been used to solve robotic control tasks. Such planners have been very successful in their various domains. The downside to such approaches is that the models and planners are highly specialised to a single control task. To change the control task, requires developing an entirely new planner. The research in my thesis focuses on the problem of specialisation in continuous, noisy and non-deterministic robotic domains by developing a more generic planner. It builds on previous research in the area, specifically using the technique of Multi-Strategy Learning. Qualitative Modelling and Qualitative Reasoning is used to provide the generality, from which specific, Quantitative controllers can be quickly learnt. The resulting system is applied to a real world robotic platform for rough terrain navigation.