Computer implemented determination method and system
Methods for providing a computer implemented medical diagnosis are provided. In one aspect, a method includes receiving an input from a user comprising at least one symptom of the user, and providing the at least one symptom as an input to a medical model. The method also includes deriving estimates of the probability of the user having a disease from the discriminative model, inputting the estimates to the inference engine, performing approximate inference on the probabilistic graphical model to obtain a prediction of the probability that the user has that disease, and outputting the probability of the user having the disease for display by a display device.
1. A method for providing a computer implemented medical diagnosis, the method comprising:
receiving an input from a user comprising at least one symptom of the user;
providing the at least one symptom as an input to a medical model comprising:
a discriminative model comprising a neural network pre-trained to approximate a probabilistic graphical model, the discriminative model being trained using samples from said probabilistic graphical model, each of the samples comprising:
a first one or more elements corresponding to each of one or more symptoms, wherein the one or more symptoms comprise the at least one symptom; and
a second one or more elements corresponding to each of one or more diseases,
wherein some observed data of the samples has been hidden to allow the discriminative model to produce data which is robust to the user providing incomplete information about their symptoms, wherein said probabilistic graphical model is a generative model comprising probability distributions and relationships between symptoms and diseases, and wherein the neural network comprises:
a plurality of input neurons comprising:
a first one or more input neurons corresponding to each of the one or more symptoms; and
a second one or more input neurons corresponding to each of the one or more diseases; and
a plurality of output neurons comprising:
a first one or more output neurons corresponding to each of the one or more symptoms; and
a second one or more output neurons corresponding to each of the one or more diseases; and
deriving an estimate, from the discriminative model, of the probability of the user having a disease of the one or more diseases, comprising:
inputting, to at least one input neuron of the first one or more input neurons, at least one value indicating that the user has the at least one symptom;
inputting, to the second one or more input neurons, one or more values indicating that the one or more diseases are unobserved; and
obtaining, from an output neuron of the second one or more output neurons, an estimate of the probability of the user having the disease; and
outputting the probability of the user having the disease for display by a display device.
2. The method according to claim 1 , wherein the neural network is a neural network comprising a plurality of sub-networks that can approximate the outputs of the probabilistic graphical model.
3. The method according to claim 1 , wherein the neural network is a single neural network that can approximate the outputs of the probabilistic graphical model.
4. The method according to claim 1 , wherein the probabilistic graphical model is a noisy-OR model.
5. The method according to claim 1 , wherein the probabilistic graphical model expresses probabilistic relationships between variables comprising diagnosis, symptoms and risk factors for medical diagnosis.
6. The method according to claim 1 , further comprising:
obtaining at least one risk factor of the user;
wherein:
each of the samples further comprises a third one or more elements corresponding to each of one or more risk factors comprising the at least one risk factor;
the plurality of input neurons further comprises a third one or more input neurons corresponding to each of the one or more risk factors;
the plurality of output neurons further comprises a third one or more output neurons corresponding to each of the one or more risk factors; and
deriving an estimate, from the discriminative model, of the probability of the user having the disease of the one or more diseases, further comprises:
inputting, to at least one input neuron of the third one or more input neurons, at least one value indicating that the user has the at least one risk factor.
7. A method for providing a computer implemented determination process for determining a probable cause from a plurality of causes, the method comprising:
receiving an input from a user comprising at least one observation;
providing the at least one observation as an input to a determination model comprising:
a discriminative model comprising a neural network pre-trained to approximate a probabilistic graphical model, the discriminative model being trained using samples from said probabilistic graphical model, each of the samples comprising:
one or more elements corresponding to each of one or more observations, wherein the one or more observations comprise the observation; and
a plurality of elements corresponding to each of the plurality of causes,
wherein some observed data of the samples has been hidden to allow the discriminative model to produce data which is robust to the user providing incomplete information about the observations, wherein the probabilistic graphical model is a generative model comprising probability distributions and relationships between observations and causes, and wherein the neural network comprises:
a first plurality of input neurons comprising:
a first one or more input neurons corresponding to each of the one or more observations; and
a second plurality of input neurons corresponding to each of the plurality of causes; and
a first plurality of output neurons comprising:
a first one or more output neurons corresponding to each of the one or more observations; and
a second plurality of output neurons corresponding to each of the plurality of causes; and
deriving estimates, from the discriminative model, of the probability of the most probable cause of the observations, comprising:
inputting, to at least one input neuron of the first one or more input neurons, at least one value indicating the at least one observation;
inputting, to the second plurality of input neurons, a plurality of values indicating that the plurality of causes are unobserved; and
obtaining, from an output neuron of the second plurality of output neurons, an estimate of the probability of the most probable cause of the at least one observation; and
outputting the probability of the most probable cause for the inputted observations for display by a display device.
8. A non-transitory carrier medium comprising computer readable code configured to cause a computer to perform a method for providing a computer implemented medical diagnosis, the method comprising:
receiving an input from a user comprising at least one symptom of the user;
providing the at least one symptom as an input to a medical model comprising:
a discriminative model comprising a neural network pre-trained to approximate a probabilistic graphical model, the discriminative model being trained using samples from said probabilistic graphical model, each of the samples comprising:
a first one or more elements corresponding to each of one or more symptoms, wherein the one or more symptoms comprise the at least one symptom; and
a second one or more elements corresponding to each of one or more diseases,
wherein some observed data of the samples has been hidden to allow the discriminative model to produce data which is robust to the user providing incomplete information about their symptoms, wherein said probabilistic graphical model is a generative model comprising probability distributions and relationships between symptoms and diseases, and wherein the neural network comprises:
a plurality of input neurons comprising:
a first one or more input neurons corresponding to each of the one or more symptoms; and
a second one or more input neurons corresponding to each of the one or more diseases; and
a plurality of output neurons comprising:
a first one or more output neurons corresponding to each of the one or more symptoms; and
a second one or more output neurons corresponding to each of the one or more diseases; and
deriving estimates, from the discriminative model, of the probability of the user having a disease of the one or more diseases, comprising:
inputting, to at least one input neuron of the first one or more input neurons, at least one value indicating that the user has the at least one symptom;
inputting, to the second one or more input neurons, one or more values indicating that the one or more diseases are unobserved; and
obtaining, from an output neuron of the second one or more output neurons, an estimate of the probability of the user having the disease; and
outputting the probability of the user having the disease for display by a display device.
9. A non-transitory carrier medium comprising computer readable code configured to cause a computer to perform a method for providing a computer implemented determination process for determining a probable cause from a plurality of causes, the method comprising:
receiving an input from a user comprising at least one observation;
providing the at least one observation as an input to a determination model comprising:
a discriminative model comprising a neural network pre-trained to approximate a probabilistic graphical model, the discriminative model being trained using samples from said probabilistic graphical model, each of the samples comprising:
one or more elements corresponding to each of one or more observations, wherein the one or more observations comprise the observation; and
a plurality of elements corresponding to each of the plurality of causes,
wherein some observed data of the samples has been hidden to allow the discriminative model to produce data which is robust to the user providing incomplete information about the observations, wherein the probabilistic graphical model is a generative model comprising probability distributions and relationships between observations and causes, and wherein the neural network comprises:
a first plurality of input neurons comprising:
a first one or more input neurons corresponding to each of the one or more observations; and
a second plurality of input neurons corresponding to each of the plurality of causes; and
a first plurality of output neurons comprising:
a first one or more output neurons corresponding to each of the one or more observations; and
a second plurality of output neurons corresponding to each of the plurality of causes; and
deriving estimates, from the discriminative model, of the probability of the most probable cause of the observations, comprising:
inputting, to at least one input neuron of the first one or more input neurons, at least one value indicating the at least one observation;
inputting, to the second plurality of input neurons, a plurality of values indicating that the plurality of causes are unobserved; and
obtaining, from an output neuron of the second plurality of output neurons, an estimate of the probability of the most probable cause of the at least one observation; and
outputting the probability of the most probable cause for the inputted observations for display by a display device.
10. A system for providing a computer implemented medical diagnosis, the system comprising:
receiving an input from a user comprising at least one symptom;
providing the at least one symptom as an input to a medical model comprising:
a discriminative model comprising a neural network pre-trained to approximate a probabilistic graphical model, the discriminative model being trained using samples from said probabilistic graphical model, each of the samples comprising:
a first one or more elements corresponding to each of one or more symptoms, wherein the one or more symptoms comprise the at least one symptom; and
a second one or more elements corresponding to each of one or more diseases,
wherein some observed data of the samples has been hidden to allow the discriminative model to produce data which is robust to the user providing incomplete information about their symptoms, wherein the probabilistic graphical model is a generative model comprising probability distributions and relationships between symptoms and diseases, and wherein the neural network comprises:
a plurality of input neurons comprising:
a first one or more input neurons corresponding to each of the one or more symptoms; and
a second one or more input neurons corresponding to each of the one or more diseases; and
a plurality of output neurons comprising:
a first one or more output neurons corresponding to each of the one or more symptoms; and
a second one or more output neurons corresponding to each of the one or more diseases;
deriving estimates, from the discriminative model, of the probability of the user having a disease, comprising:
inputting, to at least one input neuron of the first one or more input neurons, at least one value indicating that the user has the at least one symptom;
inputting, to the second one or more input neurons, one or more values indicating that the one or more diseases are unobserved; and
obtaining, from an output neuron of the second one or more output neurons, an estimate of the probability of the user having the disease; and
outputting the probability of the user having the disease for display by a display device.
11. The system of claim 10 , further comprising a graphical processing unit.