Special Action for the new German Länder: "Combination of adaptive recursive data processing with Neural Network Classification for Diagnosis in Neonates" (Project 5433 Extension)

NEUFODI -CPL - 7534

Keywords forecasting, Germany, medical diagnosis, neural networks, technology transfer

Start Date: 01 December 1992 / Duration: 12 months / Status: completed

[ contact / participants ]

the integration of adaptive signal processing with neural networks for cerebral status monitoring thereby contributing to the development of neonatology

Objectives and Approach

The main tasks of this project will be:

• to realise the integration of the optimal choice of the adaptation factors within the training algorithm of a neural network
• to investigate the properties of such hybrid systems
• to use different types of strategies within this context

Input parameters (derived from signals) for neural networks were prepared through a special pre-process based on adaptive algorithms for estimation of time courses of spectral parameters and adaptive thresholds.

The project objectives have been successfully achieved.

CONTACT POINT

Herbert Witte
Friedrich-Schiller-Universität
Bachstraße 18
D-07743 Jena
tel: +49 3641 8223014
fax: +49 3641 25447

Participants

Coordinator
Friedrich-Schiller-Universität [D]

NEUFODI -CPL - 7534, December 1993

please address enquiries to the ESPRIT Information Desk

html version of synopsis by Nick Cook