A caring analytics system that collects and mines heterogeneous sensor network data for identifying the actions that should be taken for improving the care and quality of life of adults with special needs is discussed. Imagine! Colorado, a Colorado Medicaid service provider, provides funding for the long-term care of special needs adults. Each resident is required to have an individualized service plan (ISP) that is updated yearly. The plan is created and implemented by the care team. Sensor data can be used to create three types of outputs that are important to the care of special needs adults. An algorithm places each event in a separate subset and shows that both the number of subsets and the posterior quickly converge to the actual number of activities and the maximum a posteriori. After accumulating sufficient data, it uses the learned similarity between the special needs adults to build new individualized plans that maximize the long-term care reward. Special needs adults whose most effective plans are the same are considered to be similar. The data about these adults are grouped together to construct new individual plans when sufficient data/experiences are collected. Using the algorithm, one can derive a following confidence bound on the learned effectiveness of existing plans. Having such a confidence bound is essential, since it provides the care givers with information about the effectiveness of each plan, thereby being able to reduce ambiguity about plans and increased confidence in these plans by caregivers.
ASJC Scopus subject areas
- Hardware and Architecture
- Electrical and Electronic Engineering