Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Chemical analysis
Reexamination Certificate
2001-07-09
2004-10-19
Marschel, Ardin H. (Department: 1631)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Chemical analysis
C703S002000, C703S011000, C435S006120, C435S007100
Reexamination Certificate
active
06807492
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to knowledge tree enablement within the medical field and more particularly but not exclusively to knowledge tree enablement in the field of studying side effects of drugs, with particular reference to liver toxicity.
BACKGROUND OF THE INVENTION
Pharmaceuticals are required to undergo a demanding set of clinical trials in order to prove both that they work and that they work safely, that is without untoward side effects. Also, different pharmaceuticals interact with each other so that certain drug combinations cannot safely be prescribed. The interactions between drugs are not widely understood and there is a tendency not to prescribe in any case of doubt.
Clinical trials generally involve large numbers of patients and by their nature produce large amounts of data. The results of the trail are generally assessed and conclusions reached as to the safety and efficacy of the drug.
However, analysis of large amounts of data is difficult. Each patient or volunteer participating in the trial has a medical history and it is difficult to spot patterns in the data that relate, for example, to a particular item of medical history. Thus it is also known to use automatic methods of data mining, in which computers are used to find all the possible relationships within a data set. Such an approach is unhelpful in the case of clinical trials since many trivial relationships are found. Furthermore an enormous amount of processing power is required to run such data mining, and a large amount of human analysis is required afterwards to remove all of the trivial relationships and to concentrate on the important data
SUMMARY OF THE INVENTION
According to a first aspect of the present invention there is thus an automated modeler to modeling of an interactive system comprising at least one biological entity and at least one pharmaceutical substance, the system comprising:
a representation of states of the system,
an input, associated with the representation for allowing users to define at least one of, the states, expected relationships between the states and independent inputs to the states,
a data miner associated with the representation to operate on data taken from the system to apply the data to the states in accordance with the defined relationships and inputs, thereby to apply numerical values to the relationships and the inputs, thereby to model the interactions.
Preferably, the states include beneficial actions of the pharmaceutical substance.
Additionally or alternatively, the states include harmful actions of the pharmaceutical substance.
Preferably, the model is usable to predict harmful interactions of the pharmaceutical substance within the system.
Preferably, the interactions are between at least one biological entity and a plurality of pharmaceutical substances.
Preferably, the interactions are between at least one pharmaceutical entity and a plurality of biological entities.
Preferably, the data is clinical trial data.
Preferably the model is usable to direct a clinical trial.
Preferably, the model is usable to direct drug administration to a patient.
Preferably the model is further operable to use a second data set to calibrate the model, and a third data set to test the model.
According to a second aspect of the present invention there is provided an automated system for processing effects on liver toxicity of at least one pharmaceutical substance in application to a biological entity, the system comprising:
a representation of states of the application,
an input, associated with the representation for allowing users to define expected relationships between the states, and independent inputs to the states,
a data miner associated with the representation to operate on data taken from the application to apply the data in accordance with the defined relationships and inputs, thereby to apply numerical values to the interactions and the inputs, to model the application.
Preferably, the application comprises a plurality of pharmaceutical substances.
According to a third aspect of the present invention there is provided a system for producing likely liver toxicity as a side effect of application of a pharmaceutical substance, the system comprising:
an input device for obtaining blood levels of ALT and AST respectively,
a comparator, associated with the input device for comparing the respective levels of ALT and AST to produce a ratio of the levels, and
a predictor associated both with the input device and the comparator, for predicting, from the levels and the ratio therebetween, a likelihood of development of liver toxicity.
Preferably, the predictor is set to conclude from low ALT and AST levels and a ratio close to 1, that a likelihood of development of liver toxicity is low.
Preferably, the predictor is set to conclude from high ALT and AST levels, that a likelihood of liver toxicity is relatively high.
Preferably, the predictor is set to conclude from a ratio not close to 1, that a likelihood of liver toxicity is relatively high.
Preferably, the predictor is set to conclude from a ratio close to 1, that a likelihood of liver toxicity is relatively low.
A preferred embodiment further comprises a thresholder for setting a threshold likelihood, above which application of the pharmaceutical substance is to be discontinued.
According to a fourth aspect of the present invention there is provided a method for modeling an interaction between at least one biological system and at least one pharmaceutical substance the method comprising:
building a state diagram of the interaction,
entering at least one of inputs to the states and interactions between the states,
defining at least one output from at least one of the states,
obtaining empirical data regarding the interaction,
carrying out data mining on the empirical data to assign at least one of values to the relationships and functions to the states, thereby to obtain a quantitative model of the interaction.
Preferably, the method comprises
randomly dividing the empirical data set into at least two data sets,
performing the data mining using only one of the sets, and
testing the model using a remaining one of the sets to ensure that the data has not been overfitted.
REFERENCES:
patent: 4970658 (1990-11-01), Durbin et al.
patent: 5325466 (1994-06-01), Kornacker
patent: 5440478 (1995-08-01), Fisher et al.
patent: 5479340 (1995-12-01), Fox et al.
patent: 5483468 (1996-01-01), Chen et al.
patent: 5546507 (1996-08-01), Staub
patent: 5787425 (1998-07-01), Bigus
patent: 5860917 (1999-01-01), Comanor et al.
patent: 5862054 (1999-01-01), Li
patent: 5875285 (1999-02-01), Chang
patent: 6032146 (2000-02-01), Chadha et al.
patent: 6073138 (2000-06-01), de l'Etraz et al.
patent: 6114114 (2000-09-01), Seilhamer et al.
patent: 6134555 (2000-10-01), Chadha et al.
patent: 6240329 (2001-05-01), Sun
patent: 6249712 (2001-06-01), Boiquaye
patent: 6263255 (2001-07-01), Tan et al.
patent: WO 00/00874 (2000-01-01), None
Constructivist Foundations of Modeling—A Kantian Perspective, Marco C. Bettoni, Internat. Journal of Intelligent Systems, vol. 12, No. 8, Aug. 1998, pp 577-595.
Artificial Intelligence and Manufacturing: A Research Planning Report, Leslie D. Interrante Aug. 8, 1997, http://sigmans.cs.umn.edu/sigmanwrk/report96.htm [34 pages].
Final Model Business Case Report for the OSD CALS IWSDB Project, An MVP Joint Venture, Man Tech International Corporation, Dec. 2, 1994, Kidwell et al., http://www2.dcnicn.com/cals/iwsdb/task07/html/a024/Fmodbus1.htm [77 pages].
Fisher Yossi
Hartman Yehuda
Oren Avraham
G.E. Ehrlich (1995) Ltd.
Insyst Ltd.
Marschel Ardin H.
Smith Carolyn L.
LandOfFree
Knowledge tree medical enablement does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Knowledge tree medical enablement, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Knowledge tree medical enablement will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3307451