Systems and methods for end of life analysis

1 . A method for end of life prediction comprising:

training a machine learning model, including a decision tree, using leaf-wise decision tree growth and a training data set of historical data for patients and associated end of life data of the patients, resulting in a trained machine learning model, at least one node of the decision tree of the trained machine learning model corresponding to an age threshold;

obtaining a first plurality of input features indicating first characteristics associated with a patient;

analyzing, via the trained machine learning model, the first characteristics based on the first plurality of input features; and

estimating an end of life associated with the patient based on the analyzing via the trained machine learning model to generate an estimated end of life date;

estimating, based on the analyzing via the trained machine learning model, an active decline status or an inactive decline status; and

using an interface generating system to present one or more graphical user interfaces including:

a first indication of the estimated end of life date, and

a second indication of the active decline status or the inactive decline status.

2 . The method of claim 1 , further comprising:

determining the first plurality of input features based on electronic medical records of the patient.

3 . The method of claim 1 , further comprising:

selecting a subset of information associated with the patient based on a type of the patient; and

determining the first plurality of input features based on the subset of information.

4 . The method of claim 1 , further comprising:

generating the first plurality of input features by processing patient notes using natural language processing.

5 . The method of claim 1 ,

wherein,

the first plurality of input features indicates the first characteristics associated with the patient at a first point in time; and

the method further comprises:

receiving a second plurality of input features indicating second characteristics associated with the patient at a second point in time after the first point in time;

analyzing, via the trained machine learning model, the second characteristics based on the second plurality of input features; and

estimating a second end of life associated with the patient based on the analyzing.

6 . The method of claim 1 ,

wherein,

the first plurality of input features includes baseline input features and additional input features used for continuous monitoring.

7 . The method of claim 1 ,

wherein,

the first plurality of input features include at least one of a primary diagnosis, co-morbidity information, age, gender, and a functional assessment of the patient.

8 . An apparatus for end of life prediction comprising:

a display:

a memory; and

one or more processors coupled to the memory, the one or more processors configured to:

train a machine learning model, including a decision tree, using leaf-wise decision tree growth and a training data set of historical data for patients and associated end of life data of the patients, resulting in a trained machine learning model, at least one node of the decision tree of the trained machine learning model corresponding to an age threshold;

obtain a plurality of input features indicating first characteristics associated with a plurality of patients;

analyze, via the trained machine learning model, the characteristics based on the first plurality of input features;

estimate an end of life date associated with a patient of the plurality of patients based on analyzing via the machine learning component;

estimate, based on the analyzing via the trained machine learning model, active decline statuses or an inactive decline statuses for the plurality of patients; and

present, via the display, one or more graphical user interfaces including:

a visual indication of the end of life date, and

a plurality of visual indicators, generated based on the plurality of input features, corresponding to a plurality of care teams, individual indicators of the plurality of visual indicators show a first percentage of patients of a particular care team having the inactive decline statuses or a second percentage of patients of the particular care team having the active decline statuses.

9 . The apparatus of claim 8 ,

wherein,

the one or more processors are further configured to determine the plurality of input features based on electronic medical records of the patient.

10 . The apparatus of claim 8 ,

wherein,

the one or more processors are further configured to:

select a subset of information associated with the patient based on a type of the patient, and

determine the plurality of input features based on the subset of information.

11 . The apparatus of claim 8 ,

wherein,

the one or more processors are further configured to generate the plurality of input features by processing patient notes using natural language processing.

12 . The apparatus of claim 8 ,

wherein,

the plurality of input features are a plurality of input features,

the plurality of characteristics are a plurality of characteristics,

the plurality of input features indicates the first characteristics associated with the patient at a first point in time; and

the one or more processors are further configured to:

receive a second plurality of input features indicating second characteristics associated with the patient at a second point in time after the first point in time,

analyze, via the trained machine learning model, the second characteristics based on the second plurality of input features, and

estimate a second end of life associated with the patient based on the analyzing via the trained machine learning model.

13 . The apparatus of claim 8 ,

wherein,

the patient is a hospice patient.

14 . The apparatus of claim 8 ,

wherein,

the plurality of input features include at least one of a primary diagnosis, co-morbidity information, age, gender, and a functional assessment of the patient.

15 . A non-transitory computer-readable medium having instructions stored thereon that, when executed by one or more processors, cause the one or more processors to:

train a machine learning model, including a decision tree, using leaf-wise decision tree growth and a training data set of historical data for patients and associated end of life data of the patients, resulting in a trained machine learning model, at least one node of the decision tree of the trained machine learning model corresponding to an age threshold;

obtain a plurality of input features indicating first characteristics associated with a patient;

analyze, via the trained machine learning model, the first characteristics based on the plurality of input features; and

estimate an end of life date associated with the patient based on the analyzing via the machine learning component;

perform a note comparison using the first plurality of input features; and

present, via a display, one or more graphical user interfaces including:

a visual indication of the end of life date, and

a quality reporting interface presenting an indication of whether medical records corresponding to the first plurality of input features show signs of copying and pasting based on the note comparison.

16 . A data analysis system, comprising:

a memory; and

one or more processors coupled to the memory, the one or more processors being configured to:

train a machine learning model, including a decision tree, using leaf-wise decision tree growth and a training data set of historical data for patients and associated end of life data of the patients, resulting in a trained machine learning model, at least one node of the decision tree of the trained machine learning model corresponding to an age threshold;

obtain health information associated with at least one patient of a health facility;

analyze the health information, using the trained machine learning model, to determine a rate of health decline associated with one or more health parameters of the at least one patient;

analyze the health information, using the trained machine learning model, to determine whether the health information shows signs of copying and pasting; and

present at least one graphical user interface including:

a first visual indicator indicating whether the rate of health decline meets a threshold for the one or more health parameters; and

a second visual indicator indicating whether the health information shows signs of copying and pasting.

17 . The data analysis system of claim 16 ,

wherein,

the one or more health parameters include a plurality of health parameters,

the one or more processors are further configured to determine whether the at least one patient is in active health decline based on whether a subset of the plurality of the health parameters indicate that the rate of health decline meets the threshold and yield a determination, and

the at least one interface indicates whether the at least one patient is in active health decline based on the determination.

18 . The data analysis system of claim 16 ,

wherein,

the patient is one of multiple patients,

the one or more processors are further configured to analyze health information to determine rates of health decline associated with the multiple patients, and

the at least one interface indicates a percentage of the multiple patients that are in active health decline based on one or more of the rates of health decline meeting a threshold.

19 . The data analysis system of claim 16 ,

wherein,

the one or more processors are further configured to compare health records to identify similarities in descriptions to yield a comparison, and

the at least one interface indicates similar health records based on the comparison.

20 . The data analysis system of claim 16 ,

wherein,

the one or more processors are further configured to compare patient care to a patient care plan to yield a comparison, and

the at least one interface indicates whether the patient care plan is being adhered to based on the comparison.

21 . The data analysis system of claim 16 ,

wherein,

the one or more processors are further configured to identify discrepancies in health measurements in health records, and

the at least one interface indicates the discrepancies.

22 . The method of claim 1 , further comprising:

using, by the trained machine learning model, one or more key features weighed more than other features to reduce a dimensionality of the decision tree, the one or more key features corresponding to a patient type.

23 . The method of claim 22 , wherein:

the patient type an Alzheimer patient type or a cancer patient type, and the one or more key features include:

a mid-upper arm circumference value as a first key feature; and

a blood pressure value as a second key feature.