Title:
SYSTEMS AND METHODS FOR AUTOMATIC AND INCREMENTAL LEARNING OF PATIENT STATES FROM BIOMEDICAL SIGNALS
Document Type and Number:
WIPO Patent Application WO/2005/109334
Kind Code:
A2
Abstract:
Given a record of instruments values over time, a user can mark the record t select
the values of particular instruments during particular time ranges (Fig. 4 item
402). They can further indicate the events (states) associated with those values
and time ranges (Fig. 4 item 404). These markings define the topology of the PNN.
The selected instruments define the input nodes of te PNN and the event(s) detected,
define theclass nodes wherein each event has a cossesponding positive class
node and a corresponding negative class node. Upon construction the PNN, training
may be added to further refine the knowledge of the neural network in a time efficient
manner (Fig. 4 item 406). Because the optimal value of sigma varies little over
training set sizes, training cases may be incrementally added to the PNN, further
adding to its recognition capabilities, without having to train the PNN on the
new cases or re-train on the old cases(Fig. 4 item 410). As such, patient specific
neural networks may be created in a time and cost efficient manner.
More Like This:
Inventors:
WILSON SCOTT B (US)
Application Number:
PCT/US2005/012983
Publication Date:
November 17, 2005
Filing Date:
April 15, 2005
Export Citation:
Assignee:
PERSYST DEV CORP (US)
WILSON SCOTT B (US)
WILSON SCOTT B (US)
International Classes:
A61B5/0476; G06E1/00; G06E3/00; G06F15/18; G06F19/00; G06G7/00; G06N3/02; G06N3/08; (IPC1-7): G06N3/08; G06E1/00; G06E3/00; G06F15/18; G06G7/00; G06N3/02
Foreign References:
US5943661A | 1999-08-24 | |||
US6282305B1 | 2001-08-28 | |||
US5559929A | 1996-09-24 | |||
US20010014776A1 | 2001-08-16 | |||
US6092059A | 2000-07-18 | |||
US5857179A | 1999-01-05 | |||
US6336119B1 | 2002-01-01 | |||
US6324532B1 | 2001-11-27 | |||
US20020007237A1 | 2002-01-17 | |||
US6157921A | 2000-12-05 |
Attorney, Agent or Firm:
SIMPSON, Amy, E. (Janofsky & Walker LLPP.O. Box 91909, San Diego CA, US)
Download PDF: