Progression profile prediction

1 . A method of predicting a progression of a condition, the method comprising:

obtaining measurement data relating to at least one measurement on a subject up to a particular time point, wherein the data is generated based on an output of a sensor configured to perform the at least one measurement in respect of the subject, wherein the measurement data comprises at least one N-dimensional input dataset of pixels or voxels associated with the subject, generated by an imaging device, wherein Nis a positive integer value, wherein the N-dimensional input dataset is an at least two-dimensional image dataset associated with the subject;

generating, for each of a plurality of the pixels or voxels within the N-dimensional input dataset, at least one parameter of a parameterized time-dependent function, using a trained model and based on the measurement data, wherein the parameterized time-dependent function is dependent on a continuous time value, and wherein the parameterized time-dependent function is indicative of a predicted progression of a condition of the subject over time after the particular time point, for each of the plurality of the pixels or voxels separately.

2 . The method of claim 1 , further comprising evaluating the parameterized time-dependent function using the at least one parameter, for at least one time point after the particular time point.

3 . A method of training a model to predict a progression of a condition, the method comprising

obtaining training data comprising measurement data relating to at least one measurement on at least one subject, wherein the measurement data is based on an output of a sensor configured to perform the at least one measurement in respect of the subject, wherein the measurement data comprises at least one N-dimensional input dataset of pixels or voxels associated with the subject, generated by an imaging device, wherein N is a positive integer value, wherein the N-dimensional input dataset is an at least two-dimensional image dataset associated with the subject, the training data further comprising, for each subject, at least one time point associated with the subject, and information indicative of a condition of the subject at the at least one time point;

generating, for each of a plurality of the pixels or voxels within the N-dimensional input dataset, at least one parameter of a parameterized time-dependent function, wherein the parameterized function is dependent on a continuous time value, using a model and based on the measurement data of a certain subject of the at least one subject, wherein the parameterized time-dependent function is indicative of a predicted progression of the condition of the certain subject over time, for each of the plurality of the pixels or voxels separately;

evaluating the parameterized time-dependent function using the at least one parameter, for the at least one time point associated with the certain subject, to obtain a predicted condition of the subject at the at least one time point;

comparing the predicted condition of the certain subject at the at least one time point associated with the certain subject to the information in the training data indicative of the condition of the certain subject at the at least one time point associated with the certain subject, to obtain a comparison result; and

updating the model based on the comparison result.

4 . The method of claim 3 , wherein

at least one first subject of the at least one subject has a first set of at least one time point associated therewith,

at least one second subject of the at least one subject has a second set of at least one time point associated therewith, and

at least one time point in the first set is different from each time point in the second set.

5 . The method of claim 3 , wherein the at least one parameter is indicative of a time point when the condition will change or a speed at which the condition will change.

6 . The method of claim 5 , wherein evaluating the parameterized time-dependent function comprises applying a threshold to the parameter that is generated using the model, wherein the threshold depends on the time point at which the parameterized time-dependent function is evaluated.

7 . The method of claim 3 , wherein the at least one parameter comprises at least one coefficient of a term of the parameterized time-dependent function.

8 . The method of claim 7 , wherein the parameterized time-dependent function comprises a Fourier series or a Taylor series.

9 . The method of claim 3 , wherein the model comprises a convolutional neural network.

10 . An apparatus for training a model to predict a progression of a condition, the apparatus comprising

an input configured to receive training data comprising measurement data relating to at least one measurement on at least one subject, wherein the measurement data is based on an output of a sensor configured to perform the at least one measurement in respect of the subject, wherein the measurement data comprises at least one N-dimensional input dataset of pixels or voxels associated with the subject, generated by an imaging device, wherein N is a positive integer value, wherein the N-dimensional input dataset is an at least two-dimensional image dataset associated with the subject, the training data further comprising, for each subject, at least one time point associated with the subject, and information indicative of a condition of the subject at the at least one time point;

a storage media for storing a model; and

a processor system configured to cause the apparatus to:

generate, for each of a plurality of the pixels or voxels within the N-dimensional input dataset, at least one parameter of a parameterized time-dependent function, wherein the parameterized time-dependent function is dependent on a continuous time value, using the model and based on the measurement data of a certain subject of the at least one subject, wherein the parameterized time-dependent function is indicative of a predicted progression of the condition of the certain subject over time, for each of the plurality of the pixels or voxels separately,

evaluate the parameterized time-dependent function using the at least one parameter, for the at least one time point associated with the certain subject, to obtain a predicted condition of the subject at the at least one time point,

compare the predicted condition of the certain subject at the at least one time point associated with the certain subject to the information in the training data indicative of the condition of the certain subject at the at least one time point associated with the certain subject, to obtain a comparison result, and

update the model based on the comparison result.

11 . The method of claim 1 , wherein the at least one parameter is indicative of a time point when the condition will change or a speed at which the condition will change.

12 . The method of claim 11 , wherein evaluating the parameterized time-dependent function comprises applying a threshold to the parameter that is generated using the model, wherein the threshold depends on the time point at which the parameterized time-dependent function is evaluated.

13 . The method of claim 1 , wherein the at least one parameter comprises at least one coefficient of a term of the parameterized time-dependent function.

14 . The method of claim 13 , wherein the parameterized time-dependent function comprises a Fourier series or a Taylor series.

15 . The method of claim 1 , wherein the model comprises a convolutional neural network.