阅读说明：本技术 基于树的数据探索和数据驱动方案 (Tree-based data exploration and data-driven scheme ) 是由 C·刘 A·E·卡特吉 于 2019-10-04 设计创作，主要内容包括：公开了一种向医师提供治疗建议以治疗患者的方法。该方法包括：从与患者数据存储库进行通信的处理器,基于来自适用于患者的患者数据存储库的所选患者人口统计信息的组合和适合于治疗患者的多个心室辅助设备(VAD)的操作参数,确定第一治疗建议,该第一治疗建议具有第一生存率并且包括使用第一VAD。然后,该方法通过在患者身上使用第一VAD获得第一信号。然后,该方法基于第一信号和第一治疗建议来确定第二治疗建议,该第二治疗建议具有第二生存率。然后,如果第二生存率高于第一生存率,则该方法向医师提供第二治疗建议。(A method of providing treatment recommendations to a physician to treat a patient is disclosed. The method comprises the following steps: from a processor in communication with a patient data store, a first therapy recommendation is determined based on a combination of selected patient demographic information from the patient data store applicable to the patient and operating parameters of a plurality of Ventricular Assist Devices (VADs) suitable for treating the patient, the first therapy recommendation having a first survival rate and comprising using the first VAD. The method then obtains a first signal by using a first VAD on the patient. The method then determines a second treatment recommendation based on the first signal and the first treatment recommendation, the second treatment recommendation having a second survival rate. The method then provides a second treatment recommendation to the physician if the second survival rate is higher than the first survival rate.)

1. A method of providing a treatment recommendation to a physician to treat a patient, the method comprising:

determining, from a processor in communication with a patient data store, a first therapy recommendation based on a combination of selected patient demographic information from the patient data store applicable to the patient and operating parameters of a plurality of Ventricular Assist Devices (VADs) suitable for treating the patient, the first therapy recommendation having a first survival rate and comprising using a first VAD;

treating the patient with the first VAD;

obtaining a first signal from a controller;

determining, by the processor, a second treatment recommendation based on the first signal and the first treatment recommendation, the second treatment recommendation having a second survival rate; and

providing, by the processor, the second treatment recommendation to the physician if the second survival rate is higher than the first survival rate.

2. The method of claim 1, further comprising:

if the second survival rate is not greater than the first survival rate, notifying the physician to continue treating the patient with the first VAD.

3. The method according to any of the preceding claims, wherein each VAD comprises at least one sensor for providing the first signal to the controller.

4. The method of any one of the preceding claims, wherein the first signal comprises information relating to the patient's vitality.

5. The method of any one of the preceding claims, wherein the first signal comprises at least one of: mean arterial pressure MAP, left ventricular pressure LVP, left ventricular end diastolic pressure LVEDP, pulmonary artery wedge pressure PAWP, pulmonary capillary wedge pressure PCWP, pulmonary artery occlusion pressure PAOP.

6. The method of any of the preceding claims, further comprising:

treating the patient with the second treatment recommendation.

7. The method according to any of the preceding claims, wherein the VAD comprises at least one of:a pump, an extracorporeal membrane oxygenation pump, namely an ECMO pump, a balloon pump and a Schwann-Gantz catheter.

8. The method of claim 7, wherein theThe pump includes any one of:a pump,A pump,A pump,Pump andand (4) a pump.

9. The method of claim 1, wherein the second treatment recommendation comprises: continuing to use the first VAD from the first therapy recommendation in addition to a second VAD.

10. The method of any one of claims 1 to 8, wherein the second treatment recommendation comprises treating the patient with at least one of: the first VAD, the second VAD and no VAD.

11. The method of any one of the preceding claims, wherein the patient is in cardiogenic shock.

12. The method of any of the preceding claims, wherein the first treatment recommendation is determined by a predictive model executed by the processor.

13. The method of claim 12, wherein the predictive model is based on a machine learning algorithm, the machine learning algorithm comprising any of: bagging and random forest algorithms, logistic regression algorithms, classification decision tree algorithms, deep learning algorithms, naive bayes algorithms, and support vector machine algorithms.

14. The method of any one of the preceding claims, wherein the types of patient demographic information include: gender, age, region, duration of support, instructions for use, and insertion site.

15. The method of any of the preceding claims, further comprising:

displaying the survival rate for each available VAD; and

identifying the VAD with the highest survival rate.

16. The method of any of the preceding claims, further comprising:

displaying the combination of selected patient demographic information for determining the survival rate using a branched tree representation.

17. The method according to any of the preceding claims, wherein the survival rate comprises a probability of survival of patients belonging to the combination of selected patient demographics when treated with VAD.

18. The method according to any one of the preceding claims, wherein the patient data repository comprises an acute myocardial infarction cardiogenic shock database (AMICS) database or a high risk percutaneous coronary intervention database (high risk PCI database).

19. A method of providing a treatment recommendation to a physician to treat a patient, the method comprising:

obtaining, by a processor, patient data from a patient data repository, the data being stored in the repository according to patient demographic information;

determining, by the processor, at least one Ventricular Assist Device (VAD) suitable for treating the patient;

determining, by the processor, for a combination of selected patient demographic information applicable to the patient, a survival rate for each suitable VAD based on data from the patient data store using a predictive model;

providing a suggested first VAD associated with the highest survival rate to a controller; and

treating the patient by the physician using the proposed first VAD.

20. The method of claim 19, further comprising:

providing the physician with the survival rate of each suitable VAD for all combinations of selected patient demographic information applicable to the patient.

21. The method of any of claims 19 to 20, further comprising:

providing to the physician a survival rate without using a VAD for each combination of selected patient demographic information applicable to the patient.

22. The method of any of claims 19 to 21, wherein the first VAD comprises at least one of:a pump, an extracorporeal membrane oxygenation pump, namely an ECMO pump, a balloon pump and a Schwann-Gantz catheter.

23. The method of claim 22, wherein theThe pump includes any one of:a pump,A pump,A pump,Pump andand (4) a pump.

24. The method of any one of claims 19 to 23, wherein the patient statistics include: age, sex, region, year of implantation, support equipment, duration of support, insertion site, and ejection fraction.

25. The method of any of claims 19 to 24, wherein the predictive model uses a machine learning algorithm to determine the survival rate.

26. The method of any of claims 19 to 25, wherein the machine learning algorithm comprises any of: bagging and random forest algorithms, logistic regression algorithms, classification decision tree algorithms, deep learning algorithms, naive bayes algorithms, and support vector machine algorithms.

27. The method of any one of claims 19 to 26, wherein the combination of selected patient demographic information follows a tree model.

28. The method of claim 27, wherein the tree model has an order of any one of: two, three, four, five, and six.

29. The method of any of claims 19 to 28, further comprising:

treating the patient with the first treatment recommendation.

30. The method of any of claims 19 to 29, further comprising:

displaying the survival rate for each available VAD; and

identifying the VAD with the highest survival rate.

31. The method of any of claims 19 to 30, further comprising:

displaying the combination of selected patient demographic information for determining the survival rate using a branched tree representation.

32. The method according to any one of claims 19 to 31 wherein the survival rate comprises a probability of survival of patients belonging to the combination of selected patient demographics when treated with a VAD.

33. The method according to any one of claims 19 to 32, wherein the patient data repository comprises an acute myocardial infarction cardiogenic shock database (AMICS database) or a high risk percutaneous coronary intervention database (high risk PCI database).

34. A system for providing a treatment recommendation to a physician for treating a patient, the system comprising:

at least one Ventricular Assist Device (VAD) including a sensor;

a processor in communication with the VAD, the processor configured to communicate with an acute myocardial infarction cardiogenic shock database (AMICS) or a high risk percutaneous coronary intervention database (HIC); and

a controller in communication with the VAD and the processor, the controller configured to perform the method of any of claims 1-33.

35. A system for providing a treatment recommendation to a physician for treating a patient, the system comprising:

a processor; and

a controller configured to perform the method of any one of claims 1 to 33.

36. A computer program comprising computer executable instructions which, when executed by a computing device comprising a processor and a controller, cause the computing device to perform the method of any of claims 1 to 33.

37. A non-transitory computer-readable storage medium having computer-readable code stored thereon, which, when executed by a computing device comprising a processor and a controller, causes the computing device to perform the method of any of claims 1-33.

Background

Acute and chronic cardiovascular conditions reduce quality of life and life expectancy. A variety of therapeutic modalities have been developed for treating the heart under such conditions, ranging from drugs to mechanical devices and implants. Ventricular Assist Devices (VADs), such as heart pump systems and catheter systems, are commonly used to treat the heart to provide hemodynamic support and promote healing. Some heart pump systems are inserted percutaneously into the heart and can be run in parallel with the native heart to supplement cardiac output. Such heart pump systems include those manufactured by Abiomed corporation of Danfoss (Danvers, MA), MassA family of devices.

Currently, the physician is provided by the VAD controller with a choice of a treatment plan using VAD for patients with cardiovascular conditions, and the choice is largely based on statistics of previous success of treating patients with similar conditions using this VAD. Conventional data analysis can analyze survival associated with using a VAD based on a single factor (e.g., the gender or age of the patient). As the types of myocardial conditions to which patients are predisposed develop, treatment plans that do not reasonably consider other factors affecting the patient's condition may treat the patient using a suboptimal VAD, thereby worsening the patient's condition.

Disclosure of Invention

The methods and systems described herein use a tree-based predictive model to provide physicians with treatment recommendations that are optimized for the patient. The method begins with determining, using a processor, a first therapy recommendation based on a combination of selected patient demographic information obtained from a patient data store applicable to the patient and operating parameters of a plurality of Ventricular Assist Devices (VADs) suitable for treating the patient, the first therapy recommendation having a first survival rate and including using the first VAD. The method then proceeds to obtain a first signal from the first VAD by treating the patient with a VAD. The first signal is obtained from a controller in communication with the first VAD. The processor then determines a second treatment recommendation based on the first signal and the first treatment recommendation, the second treatment recommendation having a second survival rate. The processor then determines whether the second survival rate is higher than the first survival rate and, if so, provides a second treatment recommendation to the physician.

In some embodiments, the method further comprises notifying the physician to continue treating the patient with the first VAD if the second survival rate is not greater than the first survival rate. Thus, if the processor determines that the second survival rate is equal to or less than the first survival rate, the patient continues to be treated using the first VAD. In certain embodiments, each VAD includes at least one sensor for providing a first signal to the controller. The sensor may be any input transducer configured to convert patient data into an electrical signal. In certain embodiments, the first signal includes information related to the patient's vitality, such as at least one of: for example, Mean Arterial Pressure (MAP), Left Ventricular Pressure (LVP), Left Ventricular End Diastolic Pressure (LVEDP), Pulmonary Artery Wedge Pressure (PAWP), Pulmonary Capillary Wedge Pressure (PCWP), Pulmonary Artery Occlusion Pressure (PAOP).

In certain embodiments, if it is determined that the second survival rate is greater than the first survival rate, the method further comprises treating the patient with a second treatment recommendation. The second treatment recommendation may include treating the patient using at least one of: a first VAD, a second VAD, and a no VAD. In some embodiments, the second therapy recommendation may be based on the first therapy recommendation, wherein an additional VAD may be recommended for use with the first VAD. In certain embodiments, the VAD comprises at least one of:a pump, an extracorporeal membrane oxygenation (ECMO) pump, a balloon pump, and a Swan-Ganz (Swan-Ganz) catheter.The pump includes any one of: impella (r) vaccinePump, ImpellaPump, ImpellaPump, ImpellaPump and ImpellaAnd (4) a pump. The foregoing methods are useful for treating patients with cardiogenic shock.

In some embodiments, the first treatment recommendation is determined by a predictive model executed by the processor. In some embodiments, the predictive model is based on a machine learning algorithm that includes any of: bagging and random forest algorithms, logistic regression algorithms, classification decision tree algorithms, deep learning algorithms, naive bayes algorithms, and support vector machine algorithms. In some embodiments, the predictive model uses patient demographic information applicable to the patient in its calculations. Such demographic information includes gender, age, region, duration of support, instructions for use, and insertion site. By using patient-specific demographic information, the final treatment recommendation provided to the patient will be more appropriate for each patient, thereby improving the treatment effect.

In some embodiments, the processor may display the survival rate of each available VAD; and find the VAD with the highest survival rate. In addition, the processor may use the branch tree representation to display a combination of selected patient demographic information for determining survival. Such a feature composition tree provides a visualization of features that have an impact on the proposed VAD (and associated survival) used to treat the patient for the physician. For purposes of this disclosure, survival includes the probability of survival of a patient belonging to a combination of selected patient demographics when treated with VAD.

In some embodiments, the patient data repository comprises an Acute Myocardial Infarction Cardiogenic Shock (AMICS) database or a high risk percutaneous coronary intervention (high risk PCI) database.

According to a second embodiment of the present disclosure, the method and system obtain data from a patient repository, wherein the data is stored in the repository according to patient demographic information. The method obtains data from a patient repository using a processor. The processor then determines at least one Ventricular Assist Device (VAD) suitable for treating the patient. The processor then determines the survival rate of each suitable VAD based on data from the patient data store for a combination of selected patient demographic information applicable to the patient using the predictive model. The processor then provides the proposed first VAD associated with the highest survival rate to the controller. The physician then treats the patient using the proposed first VAD.

In some embodiments, the method further provides the physician with the survival rate of each suitable VAD for all combinations of selected patient demographic information applicable to the patient. In certain embodiments, the processor also provides the physician with a survival rate without using the VAD for each combination of patient demographic information applicable to the patient. In certain embodiments, the first VAD may comprise at least one of:a pump, an extracorporeal membrane oxygenation (ECMO) pump, a balloon pump, and a swan-ganz catheter.The pump may comprise any one of: impella (r) vaccinePump, ImpellaPump, ImpellaPump, ImpellaPump and ImpellaAnd (4) a pump. The patient demographic information may include: age, sex, region, year of implantation, support equipment, duration of support, insertion site, and ejection fraction.

In some embodiments, the predictive model uses a machine learning algorithm to determine survival. The machine learning algorithm may include any one of: bagging and random forest algorithms, logistic regression algorithms, classification decision tree algorithms, deep learning algorithms, naive bayes algorithms, and support vector machine algorithms. In certain embodiments, the combination of selected patient demographic information follows a tree model. The tree model may have an order (order) of any one of: two, three, four, five, and six.

In some embodiments, the method may further include displaying a survival rate for each available VAD, and identifying the VAD with the highest survival rate. In some embodiments, the method may further include displaying a combination of selected patient demographic information for determining survival using the branched tree representation. Survival may include the probability of survival of a patient belonging to a combination of selected patient demographics when treated with VAD. In some embodiments, the patient data repository may include an Acute Myocardial Infarction Cardiogenic Shock (AMICS) database or a high risk percutaneous coronary intervention (high risk PCI) database.

According to a third embodiment of the present disclosure, a system for providing a treatment recommendation to a physician for treating a patient is provided. The system includes at least one Ventricular Assist Device (VAD) including a sensor. The system further includes a processor in communication with the VAD, the processor configured to communicate with an acute myocardial infarction cardiogenic shock (amis) repository or a high risk percutaneous coronary intervention (high risk PCI) repository. Further, the system includes a controller in communication with the VAD and the processor, the controller configured to perform the method according to any of the preceding embodiments.

According to a fourth embodiment of the present disclosure, a system for providing a treatment recommendation to a physician for treating a patient is provided. The system comprises a processor and a controller configured to perform the method according to any of the preceding embodiments.

According to a fifth embodiment of the present disclosure, there is provided a computer program comprising computer executable instructions which, when executed by a computing device comprising a processor and a controller, cause the computing device to perform the method according to any one of the preceding embodiments.

According to a sixth embodiment of the present disclosure, there is provided a non-transitory computer-readable storage medium having computer-readable code stored thereon, which, when executed by a computing device comprising a processor and a controller, causes the computing device to perform the method according to any one of the preceding embodiments.

Drawings

The foregoing and other objects and advantages will be apparent from the following detailed description taken in conjunction with the accompanying drawings, in which like reference characters refer to like parts throughout, and in which:

FIG. 1 shows an illustrative system according to an embodiment of the disclosure;

fig. 2 shows an illustrative flow diagram of a method for providing a treatment recommendation to a physician for treating a patient according to an embodiment of the present disclosure;

3A-3C show illustrative dendrograms having patient demographic factors that contribute to survival of a patient using a VAD;

FIG. 4A illustrates the accuracy of a prediction model of the order/depth of a dendrogram based on the method of the present disclosure;

FIG. 4B shows the output of the predictive model with optimized tree depth; and is

Fig. 5 shows an illustrative flow diagram of a method for providing a second treatment recommendation to a physician for treating a patient according to another embodiment of the present disclosure.

Detailed Description

Certain illustrative embodiments will be described in order to provide an overall understanding of the methods and systems described herein. Although the embodiments and features described herein are described specifically for use in connection with the survival rate of a ventricular assist device, it should be understood that all of the components and other features outlined below may be combined with each other in any suitable manner, and may be adapted and applied to other types of medical therapies having a survival rate associated therewith.

The systems and methods described herein use predictive modeling to determine the best treatment recommendation for cardiogenic shock patients. The treatment recommendation may include the use of a single Ventricular Assist Device (VAD) or multiple VADs in conjunction with each other. The predictive model extracts data and statistics from archived ventricular assist routines performed in the past. Such patient data may be stored in a patient data repository, such as, for example, an Acute Myocardial Infarction Cardiogenic Shock (AMICS) database or a high risk percutaneous coronary intervention (high risk PCI) database. The system and method uses a machine learning algorithm to predict survival when treating a patient with a VAD. To further customize the predictive model, patient demographic information and operating parameters of the VAD are also included in the model using the selected tree-based feature combinations. Here, the physician can combine any number of patient demographics and/or device features to obtain a treatment recommendation with realistic survival rates.

Additionally, real-time patient data from patients using VADs can also be fed into the predictive models to further optimize treatment recommendations. Such real-time patient data may include, but is not limited to, Mean Arterial Pressure (MAP), Left Ventricular Pressure (LVP), Left Ventricular End Diastolic Pressure (LVEDP), Pulmonary Artery Wedge Pressure (PAWP), Pulmonary Capillary Wedge Pressure (PCWP), Pulmonary Artery Occlusion Pressure (PAOP). VADs may include, but are not limited to:a pump, an extracorporeal membrane oxygenation (ECMO) pump, a balloon pump, and a swan-ganz catheter.The pump may include: impella (r) vaccinePump, ImpellaPump, ImpellaPump and ImpellaPumps, all manufactured by Abiomed corporation of danfoss (Danvers, MA), massachusetts.

Fig. 1 shows a block diagram of a system 100 for providing a first treatment recommendation 110 to a physician for treating a patient 120. The first therapy recommendation 110 includes an indication to the physician to use the most appropriate VAD for the condition of the patient 120. VADs may include, but are not limited to:a pump, an extracorporeal membrane oxygenation (ECMO) pump, a balloon pump, and a swan-ganz catheter.The pump may comprise ImpellaPump, ImpellaPump, ImpellaPump, ImpellaPump and ImpellaPumps, all manufactured by Abiomed corporation of danfoss (Danvers, MA), massachusetts. Using the predictive model, a first therapy recommendation 110 is determined based on the selected demographic information applicable to the patient 120 and the operating parameters of all VADs appropriate for the patient's condition. Here, the survival of all VADs is then determined based on the selected demographic information of the patient 120 and the operating parameters of the VADs. The VAD associated with the highest survival rate is then suggested to the physician. The indication may be made by a display of the calculation unit 130. With respect to fig. 1, the first recommendation 110 includes instructing a physician to treat the patient 120 using a first VAD having a first survival rate (SR 1).

The system 100 also includes a computing device 130, such as a laptop computer, for example, in communication with a patient data store 140. For simplicity, only the processor 135 of the computing device 130 is shown in FIG. 1. However, it will be understood that the computing device 130 also includes other components typically associated with computing devices, such as, for example, volatile memory (e.g., random access memory, RAM), non-volatile memory (e.g., read only memory, ROM), a display, and a connection bus that enables communication between these components, all of which are included in the present disclosure.

The computing device 130 includes a processor 135 that is capable of performing operations on data using the predictive model. The computing device 130 is in communication with a patient data store 140, which patient data store 140 includes patient data obtained from various medical institutions. According to certain embodiments of the present disclosure, the patient data store 140 may include an Acute Myocardial Infarction Cardiogenic Shock (AMICS) database or a high-risk PCI database compiled and maintained by a CRM, such as the salesforce. The patient data repository 140 stores data from, for example, patients with Acute Myocardial Infarction (AMI), high risk PCI patients, and cardiogenic shock patients. Patient data includes patient demographic information such as, for example, gender, age, and region. Patient data also includes data from previous treatments such as, for example, duration of support, instructions for use, insertion site, treatment device, and ejection fraction. Table 1 shows exemplary AMICS data. In some cases, the patient data store 140 also includes a database 145 of available VADs and their associated operating parameters.

The processor 135 of the processing unit 130 is also in communication with a controller 150, which controller 150 controls the operation of any VAD used to treat the patient 120. Each VAD includes a sensor that collects data from the patient when the VAD is used and transmits the data as a signal (such as SIG1 in fig. 1) to the controller 150. Such data may include, but is not limited to, Mean Arterial Pressure (MAP), Left Ventricular Pressure (LVP), Left Ventricular End Diastolic Pressure (LVEDP), Pulmonary Artery Wedge Pressure (PAWP), Pulmonary Capillary Wedge Pressure (PCWP), Pulmonary Artery Occlusion Pressure (PAOP). The first VAD of the first therapy recommendation 110 is connected to the controller 150 and transmits a first signal (SIG1) to the controller 150. In some embodiments, controller 150 comprises Automatic of Abiomed corporation of Danfoss (Danvers, MA), MassA controller (AIC). The controller 150 may be housed in a service center (serving hub)160, and the service center 160 may include other components to ensure that the respective VADs connected thereto are in an operational order. In certain embodiments, the processor 130, the processing unit 135, the controller 150, and the service center 160 may be housed in a workstation 170. The workstation 170 may further include a display (not shown) to indicate treatment recommendations to the physician.

TABLE 1 exemplary AMICS data

According to certain embodiments of the present disclosure, after the physician uses the first VAD to treat the patient 120, the controller 150 may additionally provide the physician with an indication of a second treatment recommendation. Here, the controller 150 receives the transmitted first signal SIG1 from the first VAD. The controller 150 then determines a second therapy recommendation that includes operating parameters using at least the second VAD, the first VAD, and all suitable VADs in the database 145 using a comparative model based on SIG 1. The survival of all suitable VADs is also determined, and the VAD with the highest survival is selected as the second VAD. Controller 150 also determines a second survival rate for the second VAD from AMICS database 140 (SR 2).

The controller 150 then compares the first survival rate SR1 with the second survival rate SR 2. If the second survival rate SR2 is higher than the first survival rate SR1, i.e., SR2> SR1, the physician is provided with a second recommendation via the display on the workstation 170. The second recommendation may include instructing the physician to treat patient 120 using the second VAD in place of the first VAD. The second recommendation may also include instructing the physician to use a combination of VADs selected from the database 145. The VAD used in the combination may comprise a VAD other than the first VAD, or at least one second VAD in addition to the first VAD. Further, the second recommendation may include not using the VAD at all, i.e., the second recommendation may instruct the physician to stop treating the patient with any VAD. If the second survival rate SR2 is not greater than the first survival rate SR1, i.e., SR2 ≦ SR1, the controller 150 indicates to the physician that no change should be made to the first treatment recommendation.

Fig. 2 shows a flow diagram of a method 200 of providing a first treatment recommendation to a physician to treat a cardiogenic shock patient according to an embodiment of the present disclosure. The method 200 is based on the features of the system 100 as previously described with respect to fig. 1. The method begins at step 210, where the processor 135 of the computing unit 130 accesses the patient data store 140. In some embodiments, the storage repository 140 may include AMICS or high risk PCI databases, including VAD database 145. Then, in step 220, the processor 135 determines a VAD suitable for treating the patient. Such suitability may be based on the clinical indication of the patient and the operating parameters of the VAD obtained from the patient data store 140. Accordingly, in step 220, processor 135 determines a candidate list of suitable VADs from patient data store 140 for treating the patient.

In step 230 of the method, processor 135 uses a predictive model to determine the survival rate of each VAD in the candidate list from step 220. The predictive model is based on a machine learning algorithm, which in turn includes, but is not limited to, bagging and random forest algorithms, logistic regression algorithms, classification decision tree algorithms, deep learning algorithms, naive bayes algorithms, and support vector machine algorithms, the details of which have been omitted from this disclosure for the sake of brevity.

For example, logistic regression algorithms are based on an equation that represents a predictive model using coefficients learned from training data. The representation of the model may be stored in the memory of the calculation unit 130 as a series of coefficients, each coefficient corresponding to a weight indicating the relative importance of a particular feature (e.g., a particular patient demographic information) and may be used to calculate a probability, which is then converted to the survival rate of the patient. For any number of features (Feature _ α, Feature _ β, Feature _ γ) and associated coefficients (α, β, γ), the probability can be computed as (1+ exp (-x))^-1Wherein x is equal to α × Feature _ α + β × Feature _ β + γ × Feature _ γ + ….

In another example, a decision tree algorithm uses a decision tree as a predictive model to draw conclusions about project target values from observations of the project. The tree depth may be a hyper-parameter in decision tree learning. Hyper-parameters are values that cannot be estimated from the data used in the model. Hyper-parameters are typically used to help estimate model parameters and can be adjusted for a given predictive modeling problem. The accuracy may be used as a performance indicator for the predictive model. By adjusting the hyper-parameters (such as tree depth) to determine the maximum accuracy of the decision tree, the system can provide an optimized machine learning model and thus better provide predictions (such as survival). Receiver Operating Characteristics (ROC) and area under the curve (AUC) may also be used as indicators for comparison prediction algorithms.

In step 240 of the method, the processor 135 compares the survival rates obtained for each VAD and determines the VAD with the highest survival rate. In step 250, the processor 135 provides a first treatment recommendation to the physician via an indication on a display connected to the processor to treat the patient using the VAD having the highest survival rate.

Using the tree-based data-driven approach described previously, a first treatment recommendation is provided in the method 200 of fig. 2 to tailor the first treatment recommendation to the patient's needs. By fine-tuning the treatment to the patient's demographic information, the patient may be provided with the most effective treatment plan with the best survival rate (e.g., based on historical analysis of the data in the data repository 140), thereby ensuring effective treatment for cardiogenic shock patients.

Fig. 3A illustrates a patient demographic information/feature combination tree in accordance with an embodiment of the present disclosure. The tree 300 shows how various types of features and treatments can be combined to determine the treatment with the highest predicted survival rate. The characteristics are determined from attributes of the data 310 stored in the data store 140. As can be seen from table 1, the data stored in the repository 140 may have various attributes such as gender, age, ejection fraction, type of catheter used (e.g., swan-ganz) and type of pump used (e.g., ECMO). Attributes of data related to demographic information of the patient are used in the combined tree 300 as state-level features, such as features 322 and 325. The status layer characteristic is a characteristic that is patient-related and cannot be changed, i.e. the status layer characteristic does not have an adjustable value. Examples of status layer characteristics include, but are not limited to, gender, age, and ejection fraction. The storage repository 140 also stores data relating to the types of available VADs. The features of such a VAD are used in the combined tree 300 as scheme (protocol) layer features, such as features 326-327. The protocol layer feature is a physician selectable VAD option that facilitates controlled operation of the VAD. The protocol layer characteristics have associated values that may be specified by the physician. Examples of recipe layer characteristics include, but are not limited to, rotor speed and flow rate.

The selected status layer features and the selected protocol layer features extract the patient data 310 from the data store 140 such that the extracted data can be used in a predictive model to determine a predicted survival rate 328-329 for each respective combination of the selected features 322-327 and to provide the predicted survival rate 328-329 to the physician. Predicted survival 328-329 is provided in the prediction layer in tree 300. In some embodiments, the prediction model may identify the combination of features 322-327 that gives the highest predicted survival. The combination of features 322-327 associated with the highest predicted survival rate is provided to the physician as a treatment recommendation. Treatment recommendations may be provided to the physician on the monitor. The combinations of features that make up the treatment recommendation may be presented to the physician in a feature combination tree, such as tree 310 in fig. 3A.

It will be appreciated that cardiac assist techniques have evolved over time. In addition, increasing the exposure and use of the VAD can improve its impact on the patient's condition (e.g., once the physician has better trained the use of the VAD, the impact of the use of the VAD on a particular condition patient will be seen from the patient data-e.g., the ejection fraction of a particular VAD may increase). To cater for such factors, the predictive model may specify a date range when the patient data 310 is extracted from the data store 140, in accordance with embodiments of the present disclosure. In such a case, the predictive model effectively weights the data and uses only patient data 310 from the repository 140 that fall within the specified date range. Although such data weighting is illustrated above by a date range, other factors may be considered in weighting the extracted data 310.

Fig. 3B illustrates an exemplary feature combination tree 360 for various types of patient demographics used in predictive models, according to an embodiment of the disclosure. As previously described, patient demographic data is stored in the patient data store 140 and may include, for example, gender, age, and region. The tree 360 shows the combination of the patient's gender 362 and ages 363, 364. These combinations are used for particular types of VAD (e.g. VAD)2.5 pump). Based on the data extracted from the patient data store 140, the feature 362 has values "male" and "female", and the features 363, 364 have values "50-59" and "60-69". Thus, based on the tree-based data combination shown in FIG. 3BUsing the data for the selected patient demographic data in the patient data store 140 and a machine learning algorithm (as described above), the survival rates 365 and 368 for each combination of trees 360 are shown in fig. 3B. FIG. 3B shows that a particular VAD has been implanted (e.g., the VAD has been implanted2.5 pumps) had a maximum survival rate of 84.86% in male patients ranging in age from 60 to 69 years. The data used to determine survival in fig. 3B is based on 6,392 records of 1,071+1,697+1,307+2,317 with features 362-364 in the patient data store 140. The effect on survival of a combination of features of patient data determined by the predictive model is illustrated in the example of fig. 3B.

Another example of the tree-based data-driven approach of the present disclosure is shown in the feature combination tree 370 shown in fig. 3C. Tree 370 shows a combination of three types of patient data: gender 372, ejection fraction 374, and feasibility of hemodynamic monitoring 376. The feasibility of hemodynamic monitoring depends on the type of VAD used. For example, swan-ganz catheters are known to have pressure sensors that can be used to facilitate such hemodynamic monitoring. The selected patient demographics for the combination in fig. 3C are: male, ejection fraction less than 30% and hemodynamic monitoring. Based on the tree-based data combination shown in FIG. 3C, using the data for the selected patient demographic data in the patient data store 140 and the predictive model (described above), the survival rate 378-379 for that particular combination of patient demographic information is shown in FIG. 3C. The highest survival rate using VAD capable of hemodynamic monitoring is 58.43% for male patients with ejection fraction less than 30%. Thus, the treatment recommendations for the physician are: VAD's (e.g. swan-ganz catheters) capable of hemodynamic monitoring are used on male patients with ejection fraction less than 30%. For completeness, branch 378 in tree 370 indicates survival when VAD not capable of hemodynamic monitoring is used for male patients with ejection fraction less than 30%. The data used to determine survival in fig. 3C is based on 584+ 421-1,005 records with features 372, 374, and 376 in the patient data store 140. The graphical representation of the feature composition trees 360 and 370 may also be displayed to the physician, for example on a monitor, along with the first and/or second treatment recommendations.

FIG. 4A shows a data diagram 400 for precision optimization of a predictive model using a decision tree algorithm. As previously described, the decision tree algorithm may use a hyperparameter as the tree depth. By adjusting the hyper-parameters (such as tree depth) to determine the maximum accuracy of the decision tree algorithm, the system can provide an optimized machine learning model and thus better provide predictions (such as survival). As shown in fig. 4A, the decision tree algorithm is able to capture all the noise of the training data when we change the hyper-parameter (tree depth) to a larger value. As can be seen from the line graph 410 in fig. 4A, this results in a high training score. However, at such large tree depths, the model overfitts the data and is not generalized enough. As a result, the cross-validation score worsens as seen by the line graph 420 in fig. 4A. For decision tree algorithms, if the tree is too shallow, e.g., fruit tree depth of 2 or 3, the prediction model is too simple and no correct prediction can be made, as shown in fig. 4A, where both training and cross-validation scores are low. Thus, the optimal tree depth for the decision tree algorithm is the depth that maximizes the cross-validation score and the training score. As shown in fig. 4A, this occurs at the peak of the cross-validation curve (point 430), with a tree depth of 6.

In machine learning, a false positive is the result of a machine indication that erroneously indicates a presence condition or attribute. Similarly, a false negative is the result of a machine indication that falsely identifies an absence condition or attribute. Ideally, the predictive model should predict the survival of the VAD that matches the actual survival when the VAD is used on the patient. However, as with most machine learning, there is no perfect model in real life, and therefore a trade-off between false positives and false negatives and vice versa is required when evaluating machine learning models. The ROC curve for a tree depth of 6 is shown in fig. 4B. The ROC curve 450 scans false positive rates from 0 to 100% and checks what the true positive rate is given by the model. Line plot 460 is ideal, with an area under the curve AUC of 100%, while line plot 470 is based on a stochastic model with an AUC of 50%. Any reasonable prediction model should stay between the line graph 460 and the line graph 470. For the decision tree algorithm discussed above with respect to fig. 4A, the true positive rate is shown by line plot 480 and the AUC is 87.4%, indicating good predictive ability of the predictive model when using the decision tree algorithm. For clarity, the decision tree algorithm is one of many machine learning algorithms that may be used as predictive models according to the present disclosure. The feature combination trees in fig. 3B and 3C are separate from the predictive model and provide the physician with an illustration of how to combine selected features and/or patient demographic information in the decision model.

Fig. 5 shows a flow diagram of a method 500 of providing a second treatment recommendation to a physician to treat a cardiogenic shock patient according to another embodiment of the present disclosure. The method 500 is based on the features of the system 100 as previously described with respect to fig. 1. Method 500 completes (works off) step 250 of method 200 in fig. 2, wherein a first treatment recommendation is provided to the physician, the first treatment recommendation comprising using a first VAD having a first survival rate (SR 1). The method 500 begins at step 510, where a first treatment recommendation is determined by the processor 135 at step 510. As previously described with respect to fig. 2, the processor 135 accesses the patient data store 140 and determines the VAD suitable for treating the patient. Such suitability may be based on, for example, the clinical indication of the patient and the operating parameters of the VAD obtained from the patient data store 140. The processor 135 determines a first candidate list of suitable VADs from the patient data store 140 for treating the patient and determines the survival of each VAD in the first candidate list using a predictive model. As previously described, the predictive model is based on machine learning algorithms, which in turn include, but are not limited to, bagging and random forest algorithms, logistic regression algorithms, classification tree algorithms, deep learning algorithms, naive bayes algorithms, and support vector machine algorithms. The VAD with the highest survival rate is used for the first therapy recommendation, which is referred to as the first VAD with the first survival rate (SR 1).

The physician is informed of the first therapy recommendation via a display unit connected to the processor 135 and the physician treats the patient with the first VAD (step 520). A feature combination tree (e.g., trees 300 and 350) may also be displayed. Each VAD includes a sensor that collects data from the patient when the VAD is used (step 530) and transmits the data as a signal (SIG1) to the controller 150 and processor 135. In some embodiments of the present disclosure, the collection of patient data or patient vital power, referred to as a patient vital power check, is performed at predetermined time intervals during a period in which the patient is being treated with the first VAD. As previously mentioned, such data may include, but is not limited to, Mean Arterial Pressure (MAP), Left Ventricular Pressure (LVP), Left Ventricular End Diastolic Pressure (LVEDP), Pulmonary Artery Wedge Pressure (PAWP), Pulmonary Capillary Wedge Pressure (PCWP), and Pulmonary Artery Occlusion Pressure (PAOP).

In step 540, the processor 135 determines a second candidate list of VADs suitable for treating the patient from the patient data store 140 based on the patient data contained in the SIG 1. It will be appreciated that as the patient is treated with the first VAD, the patient's vitality may change, and thus the patient data in the SIG1 may differ from the clinical indication of the patient used in determining the first treatment recommendation. Thus, the VAD in the second candidate list may be different from the VAD in the first candidate list. As with the VADs in the first candidate list, processor 135 uses the predictive model to determine the survival of each VAD in the second candidate list. The prediction model is based on machine learning algorithms that may include, but are not limited to, bagging and random forest algorithms, logistic regression algorithms, classification and regression tree algorithms, deep learning algorithms, decision tree algorithms, naive bayes algorithms, support vector machine algorithms, and vector quantization algorithms.

The VAD having the highest survival is determined, referred to as the second VAD (step 540), and in step 550, the processor 135 compares its survival (SR2) to the survival of the first VAD (SR 1). If SR2> SR1, a second VAD is used in a second treatment recommendation to the physician (step 560). The second treatment recommendation is communicated to the physician via a display unit connected to the processor 135. If SR2 ≦ SR1, the physician is not provided the second treatment recommendation, but is instructed (via the monitor) to continue using the first treatment recommendation and continue treating the patient with the first VAD as described in step 520 until the next patient vital check. After providing the second treatment recommendation to the physician, the method 500 may continue to perform the patient vital examination to further refine the treatment process, as shown in fig. 5.

Using the tree-based data-driven approach described previously, providing a second treatment recommendation in the method 500 of fig. 5 further customizes the first treatment recommendation with respect to the progress of the patient undergoing the treatment. This further customization ensures that the physician is provided with treatment recommendations with the highest survival rate for further treatment of the patient. By fine tuning the treatment to the patient's progress, the treatment of cardiogenic shock patients will be more effective.

In certain embodiments, the feature combination tree is also displayed on a monitor attached to a processor running the decision model. These feature combination trees are similar to those depicted in fig. 3B and 3C. Such a feature composition tree provides a visualization of features that have an impact on the proposed VAD used to treat the patient for the physician.

In certain embodiments, the second therapy recommendation may include using a single second VAD or multiple second VADs in combination with each other. For example, the first treatment recommendation may include the use of a balloon pump, while the second treatment recommendation may include the use of an ECMO pump in conjunction with a swan-ganz catheter. As another example, the first treatment recommendation may include the use of ImpellaThe pump, and the second treatment recommendation may include continuing to use Impella in conjunction with the balloon pumpAnd (4) a pump. In this case, the second candidate list will generate the two VADs with the highest survival. In certain embodiments of the present disclosure, the processor may be configured to determine the first n VADs with higher survival, where n ≧ 1. Then, in step 550 of method 500 in fig. 5, the survival of the n VADs is compared to the survival of the first VAD. According to some embodiments of the present disclosure, the second therapy recommendation may be to discontinue use of any VAD, as in "no" option 391 in fig. 3, whereby the "no VAD" option of the first VAD gives the highest survival.

With respect to the present disclosure, a computer-readable medium may comprise a computer-readable storage medium, which may be any tangible medium or means that can contain, or store the instructions for use by or in connection with an instruction execution system, apparatus, or device (such as a computer as defined above) for performing any of the methods described herein.

According to various embodiments of the disclosure, the computer program may be embodied in a computer program product comprising a tangible computer readable medium bearing computer program code embodied therein, the computer program code being usable with a processor to implement the above-described functions or methods.

References to "computer-readable storage medium", "computer program product", "tangibly embodied computer program", or the like, or "processor" or "processing circuitry", or the like, are to be understood as encompassing not only computers having different architectures such as single-processor architecture/multiprocessor architecture and sequencer/parallel architecture, but also specialized circuits such as field-programmable gate arrays, FPGAs, application-specific circuits, ASICs, signal processing devices, and other devices. References to computer program, instructions, code etc. should be understood to mean software for programmable processor firmware, programmable content such as instructions for a hardware device, e.g., for a processor, or configured settings or configuration settings for a fixed-function device, gate array, programmable logic device, etc.

By way of example, and not limitation, such "computer-readable storage media" may represent non-transitory computer-readable storage media that may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, flash memory, or any other medium that may be used to store desired program code in the form of instructions or data structures and that may be accessed by a computer. Also, any connection is properly termed a "computer-readable medium". For example, if the instructions are transmitted from a website, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. However, it should be understood that "computer-readable storage media" and data storage media do not include connections, carrier waves, signals, or other transitory media, but are instead directed to non-transitory tangible storage media. Disk and disc, as used herein, includes Compact Disc (CD), laser disc, optical disc, Digital Versatile Disc (DVD), floppy disk and blu-ray disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of "computer-readable media".

The instructions may be executed by one or more processors, such as one or more Digital Signal Processors (DSPs), general purpose microprocessors, Application Specific Integrated Circuits (ASICs), field programmable logic arrays (FPGAs), or other equivalent integrated or discrete logic circuitry. Thus, the term "processor" as used herein may refer to any of the foregoing structure or any other structure suitable for implementing the techniques described herein. In addition, in some aspects, the functions described herein may be provided within dedicated hardware and/or software modules. Furthermore, these techniques may be fully implemented in one or more circuits or logic elements.

If desired, the different steps discussed herein may be performed in a different order and/or concurrently with each other. Further, one or more of the above steps may be optional or may be combined, if desired.

The foregoing is merely illustrative of the principles of the present disclosure and the apparatus may be practiced in other ways than the described embodiments, which are presented for purposes of illustration and not of limitation. It should be understood that the methods disclosed herein, while shown for use in an automated ventricular assist system, may be applied to systems that will be used in other automated medical systems.

Variations and modifications will occur to those skilled in the art upon a review of the present disclosure. The disclosed features can be implemented in any combination and subcombination (including multiple dependent combinations and subcombinations) with one or more other features described herein. The various features described or illustrated above (including any components thereof) may be combined or integrated in other systems. Also, certain features may be omitted or not implemented.

Examples of changes, substitutions, and alterations are ascertainable by one skilled in the art and could be made without departing from the scope of the information disclosed herein. All references cited herein are incorporated by reference in their entirety and made a part of this application.