Parameter selection model using image analysis
Described herein are systems and methods of selecting treatment parameters values for treating skin lesions. A device may establish a treatment parameter selection model using a training dataset. The training dataset may include a plurality of examples. Each example may include a sample image of an example skin lesion to which a treatment is administered using an applicator. Each example may include a first label indicating success or failure of the treatment. Each example may include a second label corresponding to treatment parameters defining the treatment. The device may identify an input image. The device may determine that the input image corresponds to a skin lesion based on visual characteristics of the input image. The device may apply the treatment parameter selection model to the input image to output a recommended treatment to apply. The device may store an association between the input image and the recommended treatment.
1 . A method, comprising:
identifying, by one or more processors, a first image corresponding to a first skin lesion on a first subject;
generating, by the one or more processors, a first output identifying (i) at least one first treatment defined by one or more first parameters to apply to the first skin lesion and (ii) a likelihood of success for the at least one first treatment by applying a machine learning (ML) model to the first image, wherein the ML model is established by:
identifying a training dataset comprising a plurality of examples, each example of the plurality of examples having (i) a second image corresponding to a second skin lesion on a second subject, (ii) a first label indicating an outcome of a second treatment applied to the second skin lesion, and (iii) a second label identifying one or more second parameters defining the second treatment,
iteratively applying the ML model to the second image from each example of the plurality of examples,
generating, based on applying the ML model to the second image from each example of the plurality of examples, a respective second output identifying a third treatment defined by one or more third parameters to apply to the second image, and
updating at least one weight of the ML model based on the respective second output, the first label, and the second label of each example of the plurality of examples;
selecting, by the one or more processors, from the first output, the at least one first treatment to apply to the first skin lesion on the first subject based on the likelihood of success; and
storing, by the one or more processors, using one or more data structures, an association between the first image and the at least one first treatment.
2 . The method of claim 1 , further comprising detecting, by the one or more processors, a presence of the first skin lesion on the first subject within the first image, and
wherein generating the first output further comprises applying the ML model to the first image, responsive to detecting the presence of the first skin lesion.
3 . The method of claim 1 , further comprising:
detecting, by the one or more processors, an absence of any skin lesion on a third subject in a third image, and
providing, by the one or more processors, an indication to reacquire a fourth image of a third skin lesion on the third subject, responsive to detecting the absence of any skin lesion in the third image.
4 . The method of claim 1 , further comprising receiving, by the one or more processors, from a client device, an input identifying at least one of a trait of the first subject or an anatomical location associated with the first image; and
wherein generating the first output further comprises applying the ML model to the input received from the client device.
5 . The method of claim 1 , further comprising:
generating, by the one or more processors, a third output identifying (i) at least one second treatment defined by one or more second parameters to apply to the first skin lesion and (ii) a second likelihood of success for the at least one second treatment by applying the ML model to the first image; and
excluding, by the one or more processors, the at least one second treatment from selection, responsive to the second likelihood of success not satisfying a threshold likelihood.
6 . The method of claim 1 , wherein generating the first output further comprising generating the first output identifying (i) a plurality of first treatments each defined by one or more respective first parameters to apply to the first skin lesion and (ii) a respective likelihood of success for each of the plurality of first treatments; and
wherein selecting the at least one first treatment further comprises selecting, from the plurality of first treatments, the at least one first treatment based on a ranking of the respective likelihood for each of the plurality of first treatments.
7 . The method of claim 1 , wherein selecting the at least one first treatment further comprises selecting the least one first treatment, responsive to the likelihood of success satisfying a threshold likelihood.
8 . The method of claim 1 , wherein the one or more first parameters identify at least one of: (i) a type of treatment, (ii) an attribute of an applier for the at least one first treatment, (iii) at least one of an amount of dosage, (iv) a distance between the applier and the first skin lesion, or (v) a duration for administration of the at least one first treatment.
9 . The method of claim 1 , further comprising:
receiving, by the one or more processors, data identifying a result of the at least one first treatment applied to the first skin lesion on the first subject in accordance with the one or more first parameters; and
causing, by the one or more processors, the ML model to be updated using the data identifying the result of the at least one first treatment.
10 . The method of claim 1 , further comprising providing, by the one or more processors, for presentation on a client device, information identifying the one or more first parameters of the at least one first treatment to be applied to the first skin lesion on the first subject.
11 . A system, comprising:
one or more processors coupled with memory, configured to:
identify a first image corresponding to a first skin lesion on a first subject;
generate a first output identifying (i) at least one first treatment defined by one or more first parameters to apply to the first skin lesion and (ii) a likelihood of success for the at least one first treatment by applying a machine learning (ML) model to the first image, wherein the ML model is established by:
identifying a training dataset comprising a plurality of examples, each example of the plurality of examples having (i) a second image corresponding to a second skin lesion on a second subject, (ii) a first label indicating an outcome of a second treatment applied to the second skin lesion, and (iii) a second label identifying one or more second parameters defining the second treatment,
iteratively applying the ML model to the second image from each example of the plurality of examples,
generating, based on applying the ML model to the second image from each example of the plurality of examples, a respective second output identifying a third treatment defined by one or more third parameters to apply to the second image, and
updating at least one weight of the ML model based on the respective second output, the first label, and the second label of each example of the plurality of examples;
select, from the first output, the at least one first treatment to apply to the first skin lesion on the first subject based on the likelihood of success; and
store, using one or more data structures, an association between the first image and the at least one first treatment.
12 . The system of claim 11 , wherein the one or more processors are further configured to:
detect a presence of the first skin lesion on the first subject within the first image, and
apply the ML model to the first image, responsive to detecting the presence of the first skin lesion.
13 . The system of claim 11 , wherein the one or more processors are further configured to:
detect an absence of any skin lesion on a third subject in a third image, and
provide an indication to reacquire a fourth image of a third skin lesion on the third subject, responsive to detecting the absence of any skin lesion in the third image.
14 . The system of claim 11 , wherein the one or more processors are further configured to:
receive, from a client device, an input identifying at least one of a trait of the first subject or an anatomical location associated with the first image; and
apply the ML model to the input received from the client device.
15 . The system of claim 11 , wherein the one or more processors are further configured to:
generate a third output identifying (i) at least one second treatment defined by one or more second parameters to apply to the first skin lesion and (ii) a second likelihood of success for the at least one second treatment by applying the ML model to the first image; and
exclude the at least one second treatment from selection, responsive to the second likelihood of success not satisfying a threshold likelihood.
16 . The system of claim 11 , wherein the one or more processors are further configured to:
generate the first output identifying (i) a plurality of first treatments each defined by one or more respective first parameters to apply to the first skin lesion and (ii) a respective likelihood of success for each of the plurality of first treatments; and
select, from the plurality of first treatments, the at least one first treatment based on a ranking of the respective likelihood for each of the plurality of first treatments.
17 . The system of claim 11 , wherein the one or more processors are further configured to select the least one first treatment, responsive to the likelihood of success satisfying a threshold likelihood.
18 . The system of claim 11 , wherein the one or more first parameters identify at least one of: (i) a type of treatment, (ii) an attribute of an applier for the at least one first treatment, (iii) at least one of an amount of dosage, (iv) a distance between the applier and the first skin lesion, or (v) a duration for administration of the at least one first treatment.
19 . The system of claim 11 , wherein the one or more processors are further configured to:
receive data identifying a result of the at least one first treatment applied to the first skin lesion on the first subject in accordance with the one or more first parameters; and
cause the ML model to be updated using the data identifying the result of the at least one first treatment.
20 . The system of claim 11 , wherein the one or more processors are further configured to provide, for presentation on a client device, information identifying the one or more first parameters of the at least one first treatment to be applied to the first skin lesion on the first subject.