TECHNICAL FIELD

The present invention generally relates to automated and real-time patient care planning.

BACKGROUND ART

This section illustrates useful background information without admission of any technique described herein representative of the state of the art.

Electronic data records have become an important source of health and medical information. People seek treatment, contact medical providers, make appointments, and benefit from remote and optimized health care processes. All these activities can be performed on-line via computers and hand-held devices. Accurate information to support these processes, including determining the need and urgency for care, are important in assessing when, where, how, and how urgently people can reach appropriate treatment.

Despite the increasing availability of medical information from various sources, utilizing it for medical advice comes with a risk of mistreatment or misunderstanding the care need. This may lead into underestimating some critical conditions and also to falsely escalating other medical concerns, or both.

Before computers, the health history of a client, available resources and most significant examination needs may have been manually evaluated by a receptionist or a nurse, for example. However, in such a case, the true availability of resources has been inaccurate as last-minute changes could not have been accounted for. It has also not been possible to reliably manage availability of resources by multiple persons without use of a single booking calendar or a single medical records management system. This has prevented manually managing health care resources accurately in real time. Moreover, the examination needs have been dependent on the experience and interviewing skills of an individual receptionist or a nurse. As a result, a standardized procedure to estimate the care and examination needs and uncertainties of available and potentially missing information to assess these needs have been limited.

Manual resource management also leads into unnecessary transport of people and idling in professional service rendering. It is useful to optimize flows of people in all cases for various reasons, be thy environmental, economic, or medical.

It is an object of the invention to mitigate above identified problems or to at least provide alternative new solutions.

SUMMARY

According to a first example aspect of the invention there is provided an automatic and real-time method for patient care planning; comprising:

obtaining first data comprising: medical history information of a patient; and diagnostic model data;

using an interactive query process second to obtain second data comprising observations of current status of the patient;

producing probabilistic diagnostic data based on the first and second data; obtaining third data comprising triage model data;

based on the probabilistic diagnostic data and the third data, performing an interactive triage query process and in result producing triage information;

obtaining fourth data comprising care path data; and

producing probabilistically weighed care path information based on the fourth data and the triage information;

wherein the probabilistically weighed care path information is produced collectively accounting for each of the patient’s care need; urgency of care; and care path information.

The obtaining of the first data may be performed by a first processing engine. The first processing engine may be a diagnostic engine. The obtaining of the first data may comprise inputting session initialization data. The session initialization data may comprise an identity of the patient. The session initialization data may comprise authentication data for authorizing access to the medical history information of the patient.

The obtaining of the second data may employ one or more statistical models. The obtaining of the second data may employ deep learning. The obtaining of the second data may employ one or more natural language processing pipelines. The obtaining of the second data may also employ one or more structured data processing pipelines.

The interactive query process may comprise an exploration process. The interactive query process may employ reinforcement learning. The interactive query process may employ predictive modelling. The interactive query process may comprise performing an exploration/exploitation tradeoff of reinforcement learning.

The first processing engine may be configured to perform the obtaining of the second data. The first processing engine may be configured to perform the producing of the probabilistic diagnostic data based on the first and second data.

The obtaining of the third data may be performed by a second processing engine. The second processing engine may be a triage engine.

The obtaining of the fourth data may be performed by a third processing engine. The third processing engine may be a care path engine. The producing of the probabilistically weighed care path information may be performed by a third processing engine. The producing of the probabilistically weighed care path information may comprise estimating care need. The estimating of the care need may use a probabilistic inference framework. The probabilistic inference framework may be a Bayesian inference framework. The producing of the probabilistically weighed care path information may use the same Bayesian inference framework with the estimating of the care need or a different Bayesian inference framework.

The method may integrate information and build predictive models from a plurality of data sources. The plurality of data sources may comprise an electronic health records database. The plurality of data sources may comprise a domain-specific medical data source. The domain-specific medical data source may comprise a database comprising a network of medical information. The medical information may be or comprise symptoms and diagnoses. The domain-specific medical data source may comprise a triage assessment database. The domain-specific medical data source may comprise care path database.

A shared processing engine may operate as any one or more of the first, second and third processing engines.

The method may comprise maintaining a personalized care model. The personalized care model may comprise any one or more of following models: a diagnostic model; a triage model; and a personalized care model.

The method may be centrally performed by one or more server computers and / or cloud computing entities.

The method may further comprise causing presenting of quantified urgency estimates of possible causes and quantified indications of different care paths.

The collectively estimating of the patient’s care need; urgency; and care path may be based on online reinforcement learning and Bayesian inference.

The care path data may comprise data indicative of currently available care paths. The available care paths may be care paths available by given one or more healthcare providers.

According to a second example aspect of the invention there is provided an apparatus comprising at least one memory function and at least one processing function collectively configured to cause performing the method of the first example aspect.

The memory function may comprise one or more memory units. The memory units may be or comprise one or more random access memory units. The memory units may be co-located. Alternatively, the memory units may be distributed. The memory units may comprise one or more virtualised memory units. The memory units may comprise one or more cloud computing implemented memory units.

The processing function may comprise one or more processing units. The processing units may be or comprise one or more processors. The processing units may be co located. Alternatively, the processing units may be distributed. The processing units may comprise one or more virtualised processing units. The processing units may comprise one or more cloud computing implemented processing units.

According to a third example aspect of the invention there is provided a computer program comprising computer executable program code which when executed by at least one processor causes an apparatus at least to the method of the first example aspect.

According to a fourth example aspect of the invention there is provided a computer program product comprising a non-transitory computer readable medium having the computer program of the third example aspect stored thereon.

Some non-binding example aspects and embodiments of the present invention have been illustrated in the foregoing and in appended claims. Such aspects and embodiments are used merely to explain some possible implementations of the present invention. Some embodiments may be presented only with reference to certain example aspects of the invention. It should be appreciated that corresponding embodiments may apply to other example aspects as well.

BRIEF DESCRIPTION OF THE DRAWINGS

Some example embodiments of the invention will be described with reference to the accompanying drawings, in which:

Fig. 1 shows a schematic drawing of a system according to an embodiment of the invention;

Fig. 2 shows a block diagram of a server according to an embodiment of the invention; Fig. 3 shows an architectural diagram of a processing system of Fig. 1;

Fig. 4 shows a user interface of an example embodiment;

Fig. 5 shows a visual care path recommendation; and

Figs. 6 and 7 show a flow chart of a process of an example embodiment.

DETAILED DESCRIPTION

In the following description, like reference signs denote like elements or steps.

Fig. 1 shows a schematic drawing of a system 100 according to an embodiment of the invention and a schematic visualization of the Internet 130. The system 100 comprises one or more user devices 110 suited for man-machine interfacing, such as computers, smart phones, smart televisions or the like. The system further comprises a server 120. The user devices 110 and the server 120 are communicatively connected, which in case of Fig. 1 is implemented through the Internet 130. In another example embodiment these are partly or entirely combined devices or communicate through some other connection such as a data bus, private network or point-to-point connection. The server 120 implements a processing system 300 that is schematically illustrated in Fig. 3. Before that, let us briefly describe an example block diagram of the server 120 with reference to Fig. 2.

Fig. 2 shows a block diagram of the server 120. The server 120 comprises a processor 210, a memory 220, a non-volatile memory 222 capable of storing data while the server 120 is switched off, one or more pieces of software 230 stored in the non-volatile memory 222 (e.g., an operating system, drivers, code libraries, applications and configuration data). The server 120 further comprises one or more databases 240, an input/output function 250 for exchange of data, and also a user interface 260. It should be emphasized that any of these parts are combinable and also some of these parts may be omitted as a matter of implementation. Moreover, Fig. 2 illustrates a case in which dedicated hardware elements are used to implement respective functions such as a processor performs processing whereas generally various functions can also be implemented using one or more virtualization functions and/or cloud functions.

Fig. 3 shows an architectural diagram of a processing system 300 of an embodiment. The processing system 300 is divided into three different tiers that are applications, models and data. On the applications tier, there is drawn a diagnostics engine 310, a triage engine and a care path engine, which obtain first, second and third data directly from the data tier or via the models and output probabilistically weighed care path information.

On the models tier, there is a personalized care model 350 maintained that comprises a diagnostic model 352, a triage model 354 and a care path model 356. The personalized care model is adapted for each patient based on data tier based medical history information of the patient 360, e.g. in electronic health records (EHRs), which can be used in adapting the triage model 354 and the care path model and the personalized care model as a whole. The data tier has also a diagnoses and symptoms model 370 which can be used in some embodiments for adapting the diagnostics model 352, the triage model 354 and the personalized care model 350. Triage date 380 are used to adapt the triage model 356. Care path data 390 are used to adapt the care path model 356.

As shown in Fig. 3, the diagnostics model 352 can be used to adapt the diagnostic engine. The triage model 354 can be used to correspondingly adapt triage engine 320. The care path model 356 can be used to update the care path engine 330.

The processing system produces in an example embodiment a user interface shown in Fig. 4 on the user device 110, comprising:

a patient information panel leftmost panel 410 showing information originating from EHR database;

a second panel 420 for inputting symptoms, e.g., via an autocomplete search box; a symptom elicitation panel 430 in an upper middle box for inputting questions; a symptom observation panel 440 in the lower middle panel for showing positive and negative symptoms observed so far;

a diagnoses panel 450 in the upper right panel for showing the ranking of the diagnoses along with their urgency and probability estimates; and a care urgency estimation panel 460 in the lower right panel for showing urgency estimates.

Fig. 5 shows a visual care path recommendation.

Figs. 6 and 7 show a flow chart of an automatic and real-time process of an example embodiment, comprising:

600. obtaining first data comprising: medical history information of a patient; and diagnostic model data;

605. using an interactive query process second to obtain second data comprising observations of current status of the patient;

610. producing probabilistic diagnostic data based on the first and second data;

615. obtaining third data comprising triage model data;

620. based on the probabilistic diagnostic data and the third data, performing an interactive triage query process and in result producing triage information;

625. obtaining fourth data comprising care path data;

630. producing probabilistically weighed care path information based on the fourth data and the triage information; wherein the probabilistically weighed care path information is produced collectively accounting for each of the patient’s care need; urgency of care; and care path information;

635. employing deep learning in the obtaining of the second data;

640. using in the interactive query process that employs reinforcement learning;

645. performing, by the first processing engine, obtaining the second data and the producing of the probabilistic diagnostic data based on the first and second data;

700. performing the obtaining of the third data by a second processing engine that is a triage engine;

705. performing the obtaining of the fourth data by a third processing engine that is a care path engine; and

710. performing the producing of the probabilistically weighed care path information by a third processing engine;

715. the producing of the probabilistically weighed care path information comprises estimating care need using a Bayesian inference framework; and using in the producing of the probabilistically weighed care path information the same Bayesian inference framework with the estimating of the care need;

720. integrating information and building predictive models from a plurality of data sources comprising: an electronic health records database; a domain-specific medical data source that comprises a database comprising a network of symptoms and diagnoses; and a triage assessment database that comprises care path database;

725. operating a shared processing engine operates as any one or more of the first, second and third processing engines;

730. maintaining a personalized care model that comprises any one or more of following models: a diagnostic model; a triage model; and a personalized care model; 735. centrally performing the method by one or more server computers and / or cloud computing entities;

740. causing presenting of quantified urgency estimates of possible causes and quantified indications of different care paths;

745. basing the collectively estimating of the patient’s care need; urgency; and care path on online reinforcement learning and Bayesian inference; and

750. the care path data comprising data indicative of currently available care paths;

755. availing the available care paths are care paths by given one or more healthcare providers.

A technical effect of some example embodiments is that patient care can be automatically planned for a number of patients and different resources. Delays in identifying urgent care needs may be minimized while resource utilization may be maximized. Movement of patients and service providing personnel per amount of care rendered may be minimized. Care recommendations and interactive query process may be continuously improved by machine learning.

Various embodiments have been presented. It should be appreciated that in this document, words comprise, include and contain are each used as open-ended expressions with no intended exclusivity.

The foregoing description has provided by way of non-limiting examples of particular implementations and embodiments of the invention a full and informative description of the best mode presently contemplated by the inventors for carrying out the invention. It is however clear to a person skilled in the art that the invention is not restricted to details of the embodiments presented in the foregoing, but that it can be implemented in other embodiments using equivalent means or in different combinations of embodiments without deviating from the characteristics of the invention.

Furthermore, some of the features of the afore-disclosed embodiments of this invention may be used to advantage without the corresponding use of other features. As such, the foregoing description shall be considered as merely illustrative of the principles of the present invention, and not in limitation thereof. Hence, the scope of the invention is only restricted by the appended patent claims.