Apparatus for managing atopic dermatitis based on learning model and method therefor

1 . A method for managing atopic dermatitis by using an apparatus for managing atopic dermatitis, the method comprising:

collecting, on an hourly basis, patient data for a patient, the patient data comprising daily life factor data, biometric factor data, mental health factor data, skin status factor data, weather-related environmental factor data, and an observed atopic dermatitis severity index based on a medical record of the patient, wherein the collecting comprises:

receiving the daily life factor data, the mental health factor data, and the skin status factor data from a user terminal associated with the patient, the user terminal executing a mobile application for atopic dermatitis management,

receiving the biometric factor data and the observed atopic dermatitis severity index from a hospital server storing medical information of the patient, and

receiving the weather-related environmental factor data from a weather server providing location-based weather information in response to location information comprising a global positioning system location of the user terminal,

wherein the weather-related environmental factor data comprises one or more of humidity, temperature, fine dust, pollen concentration risk index, ultraviolet index, asthma index, and heat index;

storing the patient data collected on an hourly basis as time-series data for a plurality of patients;

training a learning model using the time-series data, the learning model configured to output an estimated atopic dermatitis severity index as a weighted sum of a plurality of variables and an error term representing a residual error between the estimated atopic dermatitis severity index and the observed atopic dermatitis severity index, wherein the plurality of variables comprises thirty variables including variables for each factor of a plurality of factors comprising a daily life factor, a biometric factor, a mental health factor, a skin status factor, and a weather-related environmental factor, wherein the training comprises iteratively adjusting a weight coefficient for each variable until the residual error falls below an error threshold;

determining for each factor, a threshold weight value selected to cause at least N variables for the factor to have respective weight coefficients greater than or equal to the threshold weight value, N being an integer greater than or equal to two;

reconstructing the learning model by selecting for each factor, variables having respective weight coefficients greater than or equal to the threshold weight value and excluding variables having weight coefficients less than the threshold weight value, wherein the reconstructing reduces the number of variables used by the learning model from thirty to a reduced number of variables that is fewer than thirty to increase computational efficiency and processing speed of a processor that executes the reconstructed learning model;

relearning the reconstructed learning model by applying the selected variables and the observed atopic dermatitis severity index to the reconstructed learning model;

predicting a future atopic dermatitis severity index by applying the patient data collected for the patient at a current time to the reconstructed learning model; and

providing a plurality of educational content through the mobile application on the user terminal in response to receiving the patient data from the patient, the plurality of educational content being provided in a stepwise sequence comprising cognitive correction content, automatic thought improvement content, and behavior correction content, wherein,

the cognitive correction content comprises an educational program that provides educational information about atopic dermatitis to correct a false perception of atopic dermatitis by the patient during a first set period,

the automatic thought improvement content comprises an educational program that is provided during a second set period after completion of the cognitive correction content, the program providing training to reduce negative thinking that negatively affects daily life and disease progression based on questionnaires and responses presented to the patient, and

the behavior correction content comprises an educational program that is provided during a third set period after completion of the automatic thought improvement content, the program providing meditation education and training for mindfulness of the patient.

2 . The method of claim 1 , wherein the training comprises applying variable values for each factor collected for the patient at a first time, and the observed atopic dermatitis severity index collected for the patient at a second time after a duration T 1 has elapsed from the first time, to the learning model.

3 . The method of claim 2 , wherein the predicting comprises applying the selected variables for each factor from the patient data collected for the patient at a current time to the reconstructed learning model to predict the future atopic dermatitis severity index for the patient at a future time after the duration T 1 has elapsed from the current time.

4 . The method of claim 1 , further comprising feeding back, to the user terminal associated with the patient, the future atopic dermatitis severity index and a treatment strategy selected fromfor a customized topical application, a maintenance therapy, and a recommendation to visit a hospital, wherein the treatment strategy corresponds to the future atopic dermatitis severity index.

5 . An apparatus for managing atopic dermatitis, the apparatus comprising:

one or more units being configured and executed by a processor using an algorithm, the algorithm which when executed, causing the processor to perform the one or more units, the one or more units comprising:

a data collection unit configured to collect, on an hourly basis, patient data for a patient, the patient data comprising daily life factor data, biometric factor data, mental health factor data, skin status factor data, weather-related environmental factor data, and an observed atopic dermatitis severity index based on a medical record of the patient wherein the data collection unit is configured to:

receive the daily life factor data, the mental health factor data, and the skin status factor data from a user terminal associated with the patient, the user terminal configured to execute a mobile application for atopic dermatitis management;

receive the biometric factor data and the observed atopic dermatitis severity index from a hospital server storing medical information of the patient; and

receive the weather-related environmental factor data from a weather server providing location-based weather information in response to location information received from the user terminal, the location information comprising a global positioning system location of the user terminal,

wherein the weather-related environmental factor data comprises one or more of humidity, temperature, fine dust, pollen concentration risk index, ultraviolet index, asthma index, and heat index;

a model learning unit configured to train a learning model using time-series data formed from patient data collected on an hourly basis for a plurality of patients, the learning model configured to output an estimated atopic dermatitis severity index as a weighted sum of a plurality of variables and an error term representing a residual error between the estimated atopic dermatitis severity index and the observed atopic dermatitis severity index, wherein the plurality of variables comprises thirty variables including variables for each factor of a plurality of factors comprising a daily life factor, a biometric factor, a mental health factor, a skin status factor, and a weather-related environmental factor, wherein the model learning unit is configured to iteratively adjust a weight coefficient for each variables until the residual error falls below an error threshold;

a determination unit configured to determine, for each factor, a threshold weight value selected to cause at least N variables for the factor to have respective weight coefficients greater than or equal to the threshold weight value, N being an integer greater than or equal to two;

a model reconstruction unit configured to reconstruct the learning model by selecting, for each factor, variables having respective weight coefficients greater than or equal to the threshold weight value and excluding variables having weight coefficients less than the threshold weight value, wherein the model reconstruction unit is configured to reduce the number of variables used by the learning model from thirty to a reduced number of variables that is fewer than thirty to increase computational efficiency and processing speed of the processor that executes the reconstructed learning model,

wherein the model learning unit is further configured to relearn the reconstructed learning model by applying the selected variables and the observed atopic dermatitis severity index to the reconstructed learning model;

a prediction unit configured to predict a future atopic dermatitis severity index by applying the patient data collected for the patient at a current time to the reconstructed learning model; and

a content providing unit configured to provide a plurality of educational content through the mobile application on the user terminal in response to receiving the patient data from the patient, the plurality of educational content being provided in a stepwise sequence comprising cognitive correction content, automatic through improvement content, and behavior correction content, wherein,

the cognitive correction content comprises an educational program that provides educational information about atopic dermatitis to correct a false perception of atopic dermatitis of the patient during a first set period,

the automatic thought improvement content comprises an educational program that is provided during a second set period after completion of the cognitive correction content and that provides training to reduce negative thinking that negatively affects daily life and disease progression based on questionnaires and responses presented to the patient, and

the behavior correction content comprises an educational program that is provided during a third set period after completion of the automatic thought improvement content and that provides meditation education and training for mindfulness of the patient.

6 . The apparatus of claim 5 , wherein the model learning unit is configured to train the learning model by applying variable values for each factor collected for the patient at a first time, and the observed atopic dermatitis severity index collected for the patient at a second time after a duration T 1 has elapsed from the first time to the learning model.

7 . The apparatus of claim 6 , wherein the prediction unit is configured to apply the selected variables for each factor from the patient data collected for the patient at a current time to the reconstructed learning model to predict the future atopic dermatitis severity index for the patient at a future time after the duration T 1 has elapsed from the current time.

8 . The apparatus of claim 5 , further comprising an alarm unit configured to feed back, to the user terminal associated with the patient, the future atopic dermatitis severity index and a treatment strategy selected from a customized topical application, a maintenance therapy, and a recommendation to visit a hospital, wherein the treatment strategy corresponds to the future atopic dermatitis severity index.