Exploration of Impact Biomechanics Using Data Mining

The exploration of impact biomechanics via data mining is investigated in this paper. The issues that are particularly pertinent to the use of data miming technology on biomechanics databases are addressed. These issues include (a) relationship between the manikin tests and human tests; (b) extension from lower impact, non-injurious conditions to high impact, injurious conditions; (c) test data versus simulation data; (d) input-output categorization; (e) input-output abstraction and representation; (f) topics for new knowledge discovery; and (g) user scenarios.

Technical treatments and considerations are made on the unique characteristics and requirements involved in the biodynamics data mining. They are (a) mixture of classification and numerical prediction; (b) isolated feature space; (c) multiple dependent variables; (d) high dimensionality; (e) algorithm and parameter selection; and (f) scalable data integration and knowledge discovery. A few numerical experiments in the development of a biodynamics data mining tool are described.