Prediction of care class by local additive reference to prototypical examples

The public long-term care insurance program for the elderly in Japan set out in 2000 toward establishing a new system whereby citizens can be assured that they will receive care and be supported by the society as a whole. The insurance program includes computer-aided certification processes to estimate the needs for nursing care for clients. In this work, we show the applicability of an adaptive local nonlinear approximation method associated with the Japanese national database for automatic inference of the care class.