Computer-implemented method for clinical decision support

Abstract

A computer-implemented method for clinical decision support is described. The method comprises obtaining, at a computing device, subject data associated with a subject, the subject data comprising: at least one vital parameter determined for the subject, and at least one laboratory parameter determined for the subject; and computing, based on the obtained subject data, at least one first indicator indicative of the appropriateness for multiple overnight hospitalization of the subject.

Description

Computer-implemented method for clinical decision supportCross Reference to Related Applications

[0001] This application claims the priority benefit of U.S. Provisional Application No. 63 / 558,475 filed February 27, 2024, and U.S. Provisional Application No. 63 / 564,305 filed March 12, 2024, which are incorporated herein by reference in their entireties.Technical Field

[0002] The present disclosure generally relates to the field of clinical decision support. In particular, the present disclosure relates to a computer-implemented method indicating multiple overnight hospitalization based on data associated with a subject, for example a patient. The present disclosure also relates to a computing device configured to carry out steps of said method, to a corresponding computer program instructing the computing device to perform steps of said method, and to a computer-readable medium storing such computer program.Background

[0003] Providing appropriate care for a patient after hospital admission is important. This in particular relates to critical care, but also to other types of care, such as inpatient / outpatient care or shorter-term vs longer-term care.

[0004] There is a desire to indicate / predict, at least to some extent, an appropriate or expected care need reliably and / or swiftly, e.g. after emergency department disposition based on clinical risk. Optionally, it may be desirable to provide for an improved method and device for handling of decompensation through disposition.Summary

[0005] This is achieved by a computational / computer-implemented method and tool / system / computing device that allows for clinical decision support with respect to at least multiple overnight hospitalization, based on data associated with the patient, individual or other subject. More specifically, this is achieved by the subject matter of the independent claim. Exemplary embodiments are recited in the dependent claims and described in the following.

[0006] An aspect of the present disclosure refers to a computer-implemented method for clinical decision support (e.g., with respect to multiple overnight hospitalization), the methodcomprises: obtaining, at a computing device, subject data associated with a subject, the subject data comprising at least one vital parameter determined for the subject, and at least one laboratory parameter determined for the subject, and computing, based on the obtained subject data, at least one first indicator (e.g., risk threshold) indicative / predictive of (e.g., the appropriateness / need / likelihood / recommendation / expectation for) multiple overnight (i.e. , at least two nights, i.e., more than a single night) hospitalization of the subject.

[0007] According to an embodiment, multiple overnight hospitalization may mean a stay / duration including at least two midnight crossings in hospital. In other words, multiple overnight hospitalization may mean a duration including at least two midnight crossings. The first indicator may, in some embodiments, be seen as a threshold / categorization of subjects for a stay of at least two nights in hospital. This may, in some embodiments, be seen as a long stay hospitalization and / or the subject / patient may be seen as a so-called inpatient.

[0008] The first indicator may represent a decompensation and disposition model. The first indicator may optionally be calculated after emergency department disposition.

[0009] According to an embodiment, the indicator indicates (e.g., the appropriateness of) multiple overnight hospitalization at the point in time of the determination of the subject data. In other words, the first indicator refers to an indication for the present point in time.Optionally, the indicator is determined repeatedly, e.g. each time when subject data is updated / changed.

[0010] A further aspect of the present disclosure refers to a computing device or system including one or more processors for data processing, wherein the computing device or system is configured to carry out steps of the method described herein.

[0011] A further aspect of the present disclosure refers to a computer program, which, when executed by one or more processors of a computing device or system, instructs the computing device or system to perform steps of the method described herein.

[0012] Yet a further aspect of the present disclosure refers to a computer-readable medium, for example a non-transitory computer-readable medium, storing such computer program.

[0013] Any disclosure presented herein with reference to one aspect of the present disclosure equally applies to any other aspect of the present disclosure, unless explicitly stated otherwise.

[0014] The present disclosure, therefore, may provide for an improved clinical decision support system, for example allowing to efficiently, reliably and accurately indicate the appropriateness / need / likelihood / recommendation / expectation for multiple overnight hospitalization.

[0015] For instance, aspects of the present disclosure may facilitate discharge of patients from hospitals at an appropriate point in time. For example, unnecessary long hospitalization predictions (e.g. leading to anxiety among patients) may be efficiently avoided or reduced. Also, certain procedures and protocols may be avoided at hospitals, which may result in a better patient experience, improved utilization of healthcare resources and cost savings, all while maintaining a high safety profile.

[0016] In particular, appropriate scheduling of diagnostic and / or therapeutic treatments for a patient compared to other patients may help to improve workflows in hospital. This may possibly also reduce the health risk of the entirety of patients, as each patient may be more likely to obtain diagnostic and / or therapeutic treatments in hospital at an appropriate point in time, e.g. earlier intervention for appropriate treatment of those patients who are at acute risk. An appropriate treatment path may be chosen for the subject, based on the indication according to a method of the disclosure. For example, the patient may be sent to the appropriate clinical department and / or may be looked after by the appropriate staff, based on the indication of the disclosure.

[0017] The present disclosure may provide for better predictability of a hospitalization duration and may thus, help to reduce e.g. administrative workload for retrospective changes / corrections (e.g., concerning insurance, bills, etc.) related to the distinction between a single overnight stay and a multiple overnight stay at hospital. The predictability of a hospitalization duration may also be helpful for patients which may need to know the absence from home more precisely, e.g. due to care, professional, family-related obligations, etc..

[0018] It is emphasized that the computer-implemented method and device described herein relates to a data processing means for clinical decision support, and does not involve in and for itself any diagnostic or therapeutic activity. The disclosure may in particular be helpful in connection with the assessment of patient decompensation and disposition. For example, itmay help health care professionals, e.g., clinicians, when they have to assess the health condition and the prospects of a patient.

[0019] As used herein, any parameter value of the subject or associated with the subject, for example any parameter value for the subject data, may be obtained based on a measurement, for example based on or using one or more samples from the patient, such as blood samples. This applies to any clinical data, subject data or patient data, such as vital signs, hematology parameters and / or pertinent laboratory results of the subject described herein as being usable to compute the at least one first indicator.

[0020] It is emphasized that none of the parameter values or parameters referred to herein is limited to a particular type of measurement to obtain or measure respective parameter value.

[0021] Optionally, the subject data may include time information for at least some of the parameter values included or indicated by the subject data. A timestamp or time information associated with the one or more parameter values may indicate one or more of a measurement time for a respective parameter value at the subject, a time of recordation of a parameter value, and / or a time of receipt of a parameter value at the computing device.

[0022] As used herein, determining the at least first indicator may include computing and / or assessing the likelihood or probability. Any reference herein to a single first indicator optionally includes a plurality of first indicators, unless stated otherwise. The same applies for any other indicators of the disclosure.

[0023] For instance, determining may generally relate to or include finding out or coming to a decision about appropriateness for a multiple overnight stay, which may optionally include a calculation, a computation and / or a reasoning for the assessment of the first indicator. Alternatively or additionally, computing the first indicator may include determining based on mathematical means; algorithms, models and / or calculations the likelihood for the appropriateness of a multiple overnight stay. In particular, the determination may be computer-aided, computer-assisted and / or computer-implemented.

[0024] As used herein, obtaining the subject data may comprise accessing said data and / or retrieving the subject data e.g. from the at least one memory of the computing device that carries out the method of the present disclosure, from the memory of another computing device, or from another remote data storage, e.g., a database, a secondary memory, a cloud storage or the like. Accordingly, in some cases, retrieving the subject data may comprisedownloading said data. Additionally or alternatively, obtaining the subject data may comprise receiving said data, e.g., from a user or a computing device different from the computing device accessing the data. The two options are not mutually exclusive. For instance, obtaining the set of patient data may comprise receiving said set of data, storing said set of data in a memory of the computer device and retrieving the set of data by accessing said memory. For instance, obtaining the subject data may comprise receiving the subject data at the computing device, for example at a data storage thereof. Alternatively or additionally, obtaining the subject data may include downloading, retrieving, and / or accessing the subject data by the computing device, e.g., at a data source remote from the computing device.

[0025] According to the method described herein, one or more first indicators indicative of appropriateness for multiple overnight stay may be computed by one or more computing devices based on the subject data. For instance, a single first indicator may be computed by one or more computing devices. Alternatively or additionally, a plurality of type indicators may be computed by one or more computing devices.

[0026] As used herein, the first indicator may relate to a quantitative and / or qualitative measure indicative, representative and / or descriptive of the appropriateness for the subject for a multiple overnight stay at hospital. For example, the first indicator may include or refer to a numerical result or score on an arbitrary scale, for example between zero and 100 or any other scale, indicative of the appropriateness / need / likelihood / recommendation for multiple overnight stay. Alternatively or additionally, the first indicator may represent a particular likelihood level or tier for the appropriateness for multiple overnight stay, such as low likelihood, medium likelihood, high likelihood. Alternatively or additionally, the first indicator may indicate whether single overnight stay or multiple overnight stay is more likely than the other.

[0027] For a classification, the computing device may comprise a classifier, for example based on one or more machine learning (ML) models, as described in more detail hereinbelow.

[0028] By providing a classification result indicating a first indicator that, based on the classification result, a subject may have, interpretation of the at least first indicator may be simplified, and erroneous interpretation may be avoided, for example when compared to numerical results or scores.

[0029] According to an embodiment, the subject data may further comprise information onthe subject’s symptoms and / or on the subject’s medical history and / or on interventions undertaken at or on the subject since hospital admission and / or the subject’s demographic affiliation. This may help to obtain a more complete picture that may help for more accurate indication of the at least one indicator or first indicator.

[0030] According to an embodiment, the subject data may further comprise at least one laboratory parameter trend based on the at least one laboratory parameter determined for the subject and measured at a plurality of points in time. In other words, values of a laboratory parameter over time indicating a trend may be included in the subject data. This may help to determine a course over time as regards the laboratory parameter and, hence, an improved accuracy of the first indicator.

[0031] According to an embodiment, the subject data may comprise at least one vital parameter trend based on the at least one vital parameter determined for and measured at a plurality of points in time. In other words, values of a vital parameter over time may be included in the subject data. This may help to determine a course over time as regards a vital parameter and, hence, an improved accuracy of the first indicator.

[0032] According to an embodiment, the method may include computing a first indicator trend comprising the at least one first indicator computed for a plurality of points in time. In other words, the first indicator over time may be determined. This may help to determine a course over time as regards the first indicator and, hence, an improved accuracy of the indication.

[0033] Optionally, the first indicator trend may be based on the at least one laboratory and / or vital parameter trend. Additionally or alternatively, the first indicator trend may be visualized, e.g. at a computing device, to further help a health care professional, e.g., a clinician, in decision making.

[0034] According to an embodiment, the method further comprises computing, based on the obtained subject data, a time period prediction for a duration of the hospitalization, e.g., number of overnight stays. This may allow for more specific predictions, which may be beneficial in terms of scheduling and overall reducing patient health risk.

[0035] According to an embodiment, the method may further comprise computing at least one second indicator indicative of the appropriateness / need / risk / likelihood for critical care of the subject, in addition to the first indicator. The second indicator may represent a triage decision model. Optionally, subjects meeting the risk-threshold for critical care may almostalways meet the risk threshold for multiple overnight hospitalization in terms of the first indicator and hospitalization in general, in terms of a third indicator.

[0036] According to an embodiment, further indicators are possible, e.g. a third indicator, optionally also a fourth indicator etc..

[0037] According to an embodiment, the method may further comprise computing at least one third indicator indicative of the appropriateness / need / risk / likelihood for hospitalization (i.e., care after emergency department disposition) of the subject. The third indicator may represent a hospitalization decision model. Subjects not meeting the risk threshold for hospitalization may be recommended for discharge. The corresponding patient may optionally be a so-called outpatient.

[0038] According to an example, the following categorizations for the outcome per indicator may exemplarily be used:Subject Data / Predictors OutcomesCritical Care - composite outcome that includes any; in- hospital mortality, intensive care unit stay within 24 hours of ED disposition, critical interventions within 24 hours of ED disposition (e.g., airway breathing support, electrical therapy, specific emergency procedures, hemodynamic support, emergency medications, acutely altered mental status); cardiovascular failure, or respiratory failureArrival ModeChief Complaint Long-Stay Hospitalization - composite outcome that Vital Signs (+ Trends) includes any within 72 hours of ED disposition; remaining in Medical History the hospital (i.e., not discharged) > 2 midnights, organ Lab Results dysfunction inclusive of cardiovascular or respiratory Interventions dysfunctionAny Hospitalization - composite outcome that includes any within 72 hours of ED disposition; remaining in the hospital (i.e., not discharged), organ dysfunction inclusive of cardiovascular or respiratory dysfunction

[0039] Optionally, subjects meeting the risk-threshold for hospitalization may not meet the threshold for multiple overnight hospitalization and / or critical care, in terms of the first and second indicators, respectively.

[0040] Optionally, according to an embodiment, subjects meeting the risk threshold for critical care (in terms of the second indicator) may almost always meet the risk threshold for multiple overnight hospitalization (in terms of the first indicator) and / or general hospitalization (in terms of the third indicator). Alternatively or additionally, subjects meeting criteria for multiple overnight hospitalization (in terms of the first indicator) may meet criteria for any hospitalization (in terms of the third indicator).

[0041] According to an embodiment, there may be a group that needs, according to the third indicator, any hospitalization, but not multiple overnight hospitalization in terms of the first indicator. Said group may optionally be regarded as short-stay subjects, under so-called outpatient observation.

[0042] According to an embodiment, at least two, optionally more (or all), indicators of the plurality of indicators are at least partially computed in parallel / simultaneously. Simultaneous computing of at least two or three or more indicators may be carried out for the sake of completeness, as it may turn out that one or more indicators are not decisive for the outstanding decision; e.g., in case of immediate need for critical care, a multiple overnight indication may turn out to be of less relevance.

[0043] According to an embodiment, based on the at least one first indicator, second indicator and / or third indicator, a composite indicator indicative of the overall (e.g., acute care) health risk of the subject may be determined. The composite indicator may indicate the composite outcome, e.g., by way of a risk score. This may help to classify the overall health risk.

[0044] According to an embodiment, the composite indicator may indicate / predict / recommend the following composite outcomes: need / appropriateness / recommendation for critical care, multiple overnight I long-stay hospitalization (e.g., as inpatient), single night / short-stay hospitalization (e.g., as an outpatient under observation), and hospital discharge (e.g., as outpatient).

[0045] According to an embodiment, the composite indicator may indicate multiple overnight hospitalization, if the first indicator indicates multiple overnight hospitalization and the second indicator does not indicate critical care, and optionally, the third indicator indicates hospitalization.

[0046] According to an embodiment, the at least one first indicator may be decisive for the categorization of the subject, e.g., as indicated by the composite indicator, namely to the effect that the composite indicator corresponds to the indication of the first indicator, if the at least one second indicator indicates low / no risk for critical care and the at least one third indicator indicates a need for hospitalization. In other words, the first indicator may determine / dictate / dominate the indication of appropriateness for multiple overnight hospitalization, if the second and / or third indicators are less or not decisive. This may in particular help to further clarify a grey zone between indications obtained via the second and / or third indicators.

[0047] According to an embodiment, for one or more indications of the first, second and third indicators, optionally for each of said indicators, a respective and individual threshold for each indicator is determined. Based on these individual thresholds, an overall outcome (in terms of categorization, health risk) may be determined. As an example for a categorization in connection with a risk level between 1 and 10, the following is shown:

[0048] The disposition to consider may then be, e.g., for a composite risk level 9: critical care; and for a composite risk level of 2: discharge from hospital.

[0049] In some embodiments, predicting the composite risk based on each of these three outcomes (i.e. , the outcome per individual indicator) in tandem may separate patients into four predicted (recommended) groups:1 . critical care2. long-stay (e.g. as 'inpatient')3. short-stay (e.g. as 'outpatient under observation')4. discharged home (e.g. as 'outpatient')

[0050] According to an embodiment, computing the at least one first indicator is based on at least one machine learning model, ML model, configured to receive and process input data associated with at least a subset of the subject data. The ML model may optionally provide as output the one or more first indicators. For instance, the one or more first indicators may be provided as classification result of the ML model.

[0051] In particular, some aspects of the present disclosure can use a predictive machine learning (ML) model to analyze relevant clinical data or subject data, such as vital signs or values, hematology parameters and / or laboratory results of a subject, to compute the at least one first indicator. The at least one ML model can, for example, comprise one or more of a trained logistic regression, a trained random forest algorithm, a trained gradient boosting algorithm, and a trained artificial neural network.

[0052] For example, the method may comprise determining the at least one indicator based on current subject data and past subject data, for example obtained within or over a time window of predetermined length. Alternatively or additionally, a predefined number of parameter values in the past (i.e., historical data) may be considered, such as the last two, three, four or more parameter values of a particular parameter.

[0053] For instance, the method may comprise determining all parameter values falling within a predetermined period of time, for example preceding the time of computation of the first indicator. This may include determining whether the timestamp of a parameter value is in the predetermined period of time. Optionally, one or more of the minimum value, the maximum value, the most recent value and / or a mean value of a plurality of parameter values of a particular parameter, and values having timestamps within the predetermined period of time may be computed by the computing device and used to compute the at least one first indicator.

[0054] In particular, at a given time point, the computing device and / or at least one ML model may use all subject data of at least one parameter, recorded in the immediate past preceding the time of computation of the at least one first indicator within a predetermined period of time, such as 12 hours, 24 hours, 2 days, 3 days or more.

[0055] According to an embodiment, the method may include saving, optionally in an electronic patient file, at least a subset of the subject data as obtained and / or at least a subset of the first, second and / or third indicators as computed and / or at least the composite indicatoras determined. The output may be (automatically) generated as a clinical note and / or may be saved in an Electronic Hospital Record (EHR). Put differently, the method may allow for integration in the EHR system.

[0056] According to an embodiment, the method comprises obtaining and / or determining contextual information associated with the determined at least one first indicator. For example, the contextual information may include information or data indicative or reflective of a computational basis leading to the determined one or more first indicators.

[0057] The contextual information may include data or information for interpreting the determined one or more indicators. Exemplary contextual information may include one or more notifications, messages or other informative means, which may optionally be provided at a user interface of the computing device. In an example, the contextual information may include an indication or information about the parameters and / or parameter values used for computing the one or more first indicators, and optionally their contribution in the computation of the one or more first indicators.

[0058] According to an embodiment, the method includes displaying, at a user interface, the at least one first, second and / or third indicators of a plurality of subjects, and / or the at least one first indicator trend determined for a subject at a plurality of points in time and / or at least the composite indicator as determined. This may help a healthcare professional, e.g., a clinician, to have a better overview of the one or more subjects (i.e., patients) by means of visualization. This may even help to automatically reduce the overall patient(s) risk by way of the visualization.

[0059] According to an embodiment, the method further comprises selecting at least one further hematology, vital and / or laboratory parameter, and determining the at least one first indicator (and / or second and / or third indicator) based on the at least one further vital and / or laboratory parameter.

[0060] The at least one laboratory (blood) parameter may be selected among the group of parameters consisting of: Mean Corpuscular Hemoglobin Concentration ( MCHC )value; Troponin (Tn) value; Red blood cell (RBC) count; White blood cell (WBC) count; C-reactive protein (CRP) value; Blood Urea Nitrogen (BUN) value; Lymphocyte count or value; Hematocrit value; D-dimer value; Hemoglobin value; and Red Cell Distribution Width( RDW )value.

[0061] The at least one vital parameter may be selected among the group of parameters consisting of: body temperature; blood pressure; oxygen saturation; heart rate; and respiratory rate.

[0062] One or more of the aforementioned parameter values may be used to compute the at least one first indicator and / or the second and / or third indicator.

[0063] According to a second aspect of the present disclosure, there is provided a computing device or system including one or more processors for data processing, wherein the computing device is configured to carry out steps of the method according to the first aspect of the present disclosure, as described hereinabove and hereinbelow.

[0064] The computing device may optionally comprise a data storage, for example storing at least a part of the subject data, data derived therefrom, the at least one first indicator and / or information or data related thereto. Alternatively or additionally, software instructions or one or more computer programs may be stored at the data storage, which when executed by one or more processors of the computing device, may instruct the computing device to perform steps of the method according to the first aspect of the present disclosure, as described hereinabove and hereinbelow.

[0065] According to an embodiment, the computing device includes one or more ML models, for example implemented in a classifier circuitry or logic of the computing device, as described in more detail hereinabove and hereinbelow.

[0066] The computing device described herein may refer to any data processing device, including a standalone computing device or server and computing networks with a plurality of inter-operating computing devices, such as a cloud computing system or server system. Alternatively or additionally, the computing device may be embodied, at least in part, as mobile device, such as a smart phone a tablet computer, a notebook or the like.

[0067] The computing device may, in an example, comprise one or more communication interfaces configured to communicate with one or more remote devices, for example an external data storage or database, and / or one or more remote computing devices. For instance, at least a part of the subject data may be received via the communication interface from an external data storage. Alternatively or additionally, the computed at least one indicator or information related thereto may be transmitted to one or more remote devices or stored at the external data storage.

[0068] A third aspect of the present disclosure relates to a computer program, which, when executed by one or more processors of a computing device, instructs the computing device to perform steps of the method according to the first aspect of the present disclosure, as described hereinabove and hereinbelow.

[0069] A fourth aspect of the present disclosure relates to a computer-readable medium, for example a non-transitory computer-readable medium, storing a computer program, which, when executed by one or more processors of a computing device, instructs the computing device to perform steps of the method according to the first aspect of the present disclosure, as described hereinabove and hereinbelow. In particular, the computer-readable medium according to the fourth aspect of the present disclosure stores the computer program according to the third aspect of the present disclosure.

[0070] These and other aspects of the disclosure will be apparent from and elucidated with reference to the appended figures, which may represent exemplary embodiments.Brief Description of the Drawings

[0071] The subject-matter of the present disclosure will be explained in more detail in the following with reference to exemplary embodiments which are illustrated in the attached drawings, wherein:

[0072] Fig. 1 shows a computing device or system for determining at least one first indicator according to an exemplary embodiment.

[0073] Fig. 2 shows a flow chart illustrating a method of determining at least one first indicator according to an exemplary embodiment.

[0074] Fig. 3 shows an illustrative, non-limiting example of a computing environment in which a machine-learning model may be trained and / or deployed, according to at least one embodiment of the present disclosure.

[0075] The figures are schematic only and not true to scale. In principle, identical or like parts are provided with identical or like reference symbols in the figures.Detailed Description of Exemplary Embodiments

[0076] Figure 1 shows a computing device or system 100, for example a clinical decision support system 100, configured to determine, for a subject (patient and / or individual), the at least one first indicator indicative of the appropriateness of multiple overnight hospitalization, i.e. stay in hospital for at least two midnights crossings.

[0077] The computing device 100 comprises a processing circuitry 110 or control circuitry 110 with one or more processors 112 for data processing. Optionally, the processing circuitry 110 or control circuitry 110 may include a classifier or a classifier circuitry. The computing device 100 further comprises at least one data storage 120 for storing data and / or software instructions. For example, at least a subset of the subject data and / or the at least one fist indicator may be stored at the data storage 120.

[0078] The exemplary computing device 100 of Figure 1 further comprises at least one communication circuitry or interface 130 for communicatively coupling the computing device 100 to one or more external data sources 200 that may optionally store data and / or provide data to the computing device 100. The communication circuitry 130 may be configured for wired or wireless communication with the at least one external data source 200. It should be noted that the computing device 100 may comprise a plurality of communication circuits 130 or interfaces 130 for communicatively coupling the computing device 100 to a plurality of different external data sources 200.

[0079] The one or more external data sources 200 may for example be associated with one or more external servers communicatively coupled to the computing device 100, for example via the Internet, a LAN connection, a wireless connection or a wired connection. For example, the computing device 100 may be communicatively couplable to a hospital information system, a laboratory information system, a server of a health care provider, or any other server or data processing device.

[0080] As discussed in detail hereinabove and hereinbelow, the computing device 100 may be configured to perform steps of the method of determining the at least one first indicator based on subject data including one vital parameter determined for the subject, and at least one laboratory parameter determined for the subject. Therein, the computing device 100 receives or obtains the subject data. Further, the computing device 100 is configured to determine, based on processing the received set of patient data, and / or compute the at least one first indicator indicative / representative of (the appropriateness / need / likelihood / recommendation for) multiple overnight hospitalization, asdescribed in detail above.

[0081] The computing device 100 may be configured to carry out the method of any of the dependent claims and / or of the optionally features disclosed above.

[0082] In particular, the computing device 100 may be configured to obtain subject data comprising information on the subject’s symptoms and / or on the subject’s medical history and / or on interventions undertaken at the subject since hospital admission and / or the subject’s demographic affiliation.

[0083] The computing device 100 may be configured to obtain the subject data comprising at least one laboratory parameter trend based on the at least one laboratory parameter determined for the subject and measured at a plurality of points in time.

[0084] The computing device 100 may be configured to obtain subject data comprising at least one vital parameter trend based on the at least one vital parameter determined for and measured at a plurality of points in time.

[0085] The computing device 100 may be configured to compute a first indicator trend comprising the at least one first indicator computed for a plurality of points in time, optionally based on the at least one laboratory and / or vital parameter trend, respectively.

[0086] The computing device 100 may be configured to compute, based on the obtained subject data, a time period prediction for a duration of the hospitalization.

[0087] The computing device 100 may be configured to compute at least one second indicator indicative of the need for critical care of the subject and / or at least one third indicator indicative of the need for inpatient care of the subject, optionally within the next 72 hours.

[0088] The computing device 100 may be configured to compute at least two, optionally three indicators of plurality of indicators are at least partially computed simultaneously.

[0089] The computing device 100 may be configured to compute, based on the at least one first indicator, second indicator and / or third indicator, a composite indicator indicative of the overall acute care health risk of the subject.

[0090] The computing device 100 may be configured to apply the at least one first indicator to determines / dictate / dominate the appropriateness for multiple overnight hospitalization, optionally the composite indicator, if the at least one second indicator indicates low risk for critical care and the at least one third indicator indicates a need for inpatient care.

[0091] The computing device 100 may be configured to compute the at least one first indicator, second indicator and / or third indicator based on at least one ML model configured to receive and process input data associated with at least a subset of the subject data.

[0092] The actual computation of the at least one first indicator may be performed based on at least one ML model of the computing device 100. The at least one ML model may comprise one or more of a trained logistic regression, a trained random forest algorithm, a trained gradient boosting algorithm, and a trained artificial neural network.

[0093] Exemplarily, the ML model comprises the XGBoost algorithm defined by Chen and Guestrin. A benefit of XGBoost algorithm is that it can deal with missing information without requiring data imputation.

[0094] Optionally, the subject data may comprise a plurality of vital and / or laboratory values determined at or associated with different times, and the computing device 100 may compute one or more of a maximum value of the plurality of values, a rate of change of the value or concentration, a most recent value and a minimum value. Further optionally, a mean value may be computed. One or more of the aforementioned calculated parameter values may be used to compute the at least one first indicator. Further optionally, a plurality of values may be included in the subject data and corresponding parameter values may be computed and used to compute the first indicator.

[0095] The at least one laboratory parameter may be selected among the group of parameters consisting of: MCHC value; Troponin value; Red blood cell count; White blood cell count; C-reactive protein value; Blood Urea Nitrogen value; Lymphocyte count or value; Hematocrit value; D-dimer value; Hemoglobin value; and RDW value.

[0096] The at least one vital parameter may be selected among the group of parameters consisting of: body temperature; blood pressure; oxygen saturation; heart rate; and respiratory rate.

[0097] Also, it is noted that a plurality of any one or more of the aforementioned parameters may be comprised in the subject data and further parameter values may be derived therefromby the computing device. In particular, the computing device 100 may be configured to determine a minimum diastolic blood pressure value among a plurality of diastolic blood pressure values and compute the at least one first indicator based thereon. Alternatively or additionally, the computing device 100 may be configured to determine a minimum respiratory rate value among a plurality of respiratory rate values and compute the at least one first indicator based thereon. Alternatively or additionally, the computing device 100 may be configured to determine a minimum body temperature value among a plurality of body temperature values and compute the at least one first indicator based thereon.

[0098] The computing device 100 may be configured to save, optionally in an electronic patient file, at least a subset of the subject data as obtained and / or at least a subset of the first, second and / or third indicators as computed. The computing device 100 can also be implemented in a remote cloud environment.

[0099] The computing device 100 may be configured to display, at a user interface, the at least one first, second and / or third indicators of a plurality of subjects, and / or the at least one first indicator trend determined for a subject at a plurality of points in time.

[0100] Figure 2 shows a flow chart illustrating a computer-implemented method of determining at least one first indicator according to an exemplary embodiment, for example using a computing device 100 as described with reference to Figure 1.

[0101] Step S1 comprises obtaining, at a computing device 100, subject data associated with a subject after admission to hospital. The subject data comprises a) at least one vital parameters determined for the subject, and b) at least one laboratory parameter determined for the subject.

[0102] Step S2 comprises computing, based on the obtained subject data with the computing device 100, at least one first indicator indicative of the appropriateness for multiple overnight hospitalization of the subject.

[0103] It is noted that one or more further optional steps, as described hereinabove in the summary part and with reference to Figure 1 , may be performed for determining the at least one first indicator and / or for further steps / features according to the disclosure.

[0104] For example, in an optional step S3, the at least one first indicator and / or information indicative of the appropriateness for multiple overnight hospitalization of the subject mayoptionally be generated and potentially displayed at the user interface 140 of the computing device 100.

[0105] Further optional steps according to the optional features, e.g. as defined in the dependent claims and / or as disclosed herein above, may be carried out.

[0106] As mentioned above, the at least one first indicator may be computed based on or using one or more ML models. The one or more ML models can be trained using historical data from the emergency department (ED), data from the electronic health record (EHR), data from health care systems, e.g. across the USA, as training data (also referred to as training subject data). An inclusion criterion can be that the ED / EHR data include at least one vital parameter and one laboratory parameter value. The data can include de-identified demographic, clinical, and diagnostic information for the patient or subject in the encounter.

[0107] For example, the training data or training subject data may be obtained from the EHR data by extracting data being associated with a particular subject and being indicative of one or more of the parameters considered as input for the one or more ML models, such as any parameter described herein to be potentially usable for computing the at least one first indicator. Each instance of patient or EHR data in the training data or training subject data may be associated with a respective known outcome in order to label the training data. For instance, the label can be a binary label wherein label 1 may equal or indicate a multiple overnight stay and label 0 may indicate a single night hospitalization, or vice versa. In an example, the ML model can be trained to provide the likelihood of subject data provided as input data being associated with single or multiple overnight stay in hospital.

[0108] The EHR data may further include an indication about one or more diseases a patient suffered from in the past. For instance, diagnosis codes, such as according to the International Statistical Classification of Diseases and Related Health Problems, so-called ICD codes, can be stored in the EHR data.

[0109] The training subject data obtained from the EHR data may optionally be cleaned for subsequent processing, for example based on removing erroneous data records. Further optionally, the training subject data may be split into a training dataset, which can be used for optimizing the one or more ML models, and a test dataset, which can be used for validation of the one or more ML models. For instance, model parameters of the one or more ML models can be tuned or adjusted using the training dataset to determine optimized model parameters, which may then be used to validate the one or more ML models using the test dataset. Theprocess can be repeated iteratively, for example until the one or more ML models satisfy a target performance on the test dataset.

[0110] In the following, an exemplary implementation of a ML model and aspects of training such model are summarized. The ML model may include one or more of a trained logistic regression, a trained random forest algorithm, a trained gradient boosting algorithm, and a trained artificial neural network.

[0111] In particular, the method described herein may be implemented at least in part in a logistic regression ML model. Alternatively, a boosting technique may be utilized, such as XGBOOST, which however, is optional only.

[0112] When trained and / or during inference of the trained ML model, the at least one vital parameter and the at least one laboratory parameter of a subject can be provided as input to the trained ML model. The at least one first indicator may then be computed and a corresponding output may be generated, for example in the form of %-appropriateness for multiple overnight hospitalization.

[0113] Figure 3 shows an illustrative, non-limiting example of a computing environment 300 in which a ML model may be trained and / or deployed, according to at least one embodiment of the present disclosure. According to at least one embodiment, a computing environment 300 described herein comprises one or more of: an untrained model 304; training data 302; training framework 306; a trained model 308; an evaluator 310; classifier 312; new input data 314; and inferences 316.

[0114] In at least one embodiment, an untrained model 304 refers to an untrained ML model - this may comprise or otherwise be in the form of a logistic regression model, a random forest algorithm, a decision tree, an artificial neural network, or other such models described throughout this disclosure. Typically, an untrained model 304 may be viewed as the starting point of a training process where the model begins with no knowledge of patterns within the data and may be initialized with random or default parameter values. In some cases, pretraining occurs and the untrained model 304 is in a pre-trained state. As was described above and will be further expanded upon below, various learning techniques may be applied to untrained model 304 to adjust the model’s parameter values to perform various tasks, such as classification tasks for improved clinical decision support.

[0115] In various embodiments, training data 302 refers to a dataset comprising labeled data (in supervised learning) that allows for a ML model (e.g., untrained model 304 or a partially trained or pre-trained model) to learn patterns within the data. In various embodiments, training data 302 is divided into two non-overlapping subsets: a training set and a validation set. The training set is used to update the model’s parameters over training. As the model is exposed to a wide range of different training data (e.g., training subject data), it will be able to learn patterns, relationships, and features and encode the learnings into the various model parameters. Conversely, a validation set refers to a distinctly separate subset of the training data 302 that is not directly used during the training phase but is instead used during validation. The validation data is used to evaluate whether a ML model, as it is being trained, is able to respond well to previously unseen input data, and can be used to fine-tune the ML model parameters, detect overfitting, and otherwise evaluate and improve the model’s performance. In some cases, an addition subset of the training data 302 is reserved as a test set for evaluating the final ML model after training has been completed.

[0116] In various embodiments, training framework 306 refers to hardware, software, or a combination thereof that manages the ML learning process. In various embodiments, training framework 306 includes ML libraries such as PyTorch or TensorFlow, which provide softwarebased tools for defining model architectures, computing gradients, optimization parameters, and so on and so forth. In various embodiments, training framework 306 is responsible for handling input training data, which can involve data loading, data preprocessing, data pipelining, and the like. In various embodiments, training framework 306 provides for supervised learning of a ML model. In supervised learning, the ML model is tasked with generating predictions and a true output (ground truth) is withheld from the model. The model’s prediction is compared against the label. A loss function may be computed to evaluate the accuracy of the ML model and / or used for model optimization.

[0117] In various embodiments, training framework 306 produces a trained model 308. Generally speaking, training can be performed by taking an initial ML model (e.g., untrained model 304) and adjusting parameters (weights) of the ML model through a training process. A training set is divided into features and labels, with features being withheld from the model during predictions. In an illustrative example of a training process, the training set is divided into mini-batches. A single iteration is performed over a mini-batch and the model’s performance is evaluated and updated. After each iteration, the ML model makes a prediction in a forward pass and a loss value is computed as the difference between the prediction and the ground truth label that was withheld. A backwards pass is then performed where gradients are computed for each model parameter and updated for the model, and the weights of it. Themodel may be iteratively trained on successive mini-batches, until the model has been exposed to all data in the training set, thereby completing one epoch of training. In various embodiments, the ML model is trained over several epochs to ensure that the model is able to learn various complex patterns within the training set.

[0118] In various embodiments, a trained model 308 is produced as the final output of the training framework 306. A ML model may be refined over multiple epochs until the loss function stabilizes, the validation accuracy stops improving, after a certain amount of time or number of epochs have been exhausted, or other such criteria. Regardless, once the trained model 308 has been produced, it can optionally be provided to an evaluator 310. In various embodiments, evaluator 310 assesses the model’s performance by using data that was unseen during training (e.g., a test set) to test the generalizability of the model. The model’s accuracy may be tested against this unseen data to ensure that the model is not subject to overfitting and is able to handle new data. If the model’s performance is unsatisfactory, it may be rejected and the training framework 306 may be tasked with producing a new model, for instance, with different training parameters and / or different training data. If trained model 308 is accepted, then the trained model may be used for inferencing.

[0119] In various embodiments, trained model 308 is provided for use by a classifier 312 for classifying or otherwise generating inferences. In various embodiments, classifier 312 obtains new input data 314 and produces inferences 316. Generally speaking, new input data will not necessarily be labeled and / or the ground truth is not necessarily known or even knowable. In at least one embodiment of the present disclosure, classifier 312 obtains new input data 314 in the form of subject data associated with a subject, the subject data comprising: at least one vital parameter determined for the subject and at least one laboratory parameter determined for the subject. In various embodiments, classifier 312 produces an inference 316 based on the obtained subject data, at least one first indicator, the inference 316 being indicative of the appropriateness for multiple overnight hospitalization of the subject.

[0120] While the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive; the invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and effected by those skilled in the art and practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims.

[0121] In the claims, the word “comprising” does not exclude other elements or steps, and the indefinite article “a” or “an” does not exclude a plurality. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. Any reference signs in the claims should not be construed as limiting the scope.

Claims

Claims1 . A computer-implemented method for clinical decision support, the method comprising: obtaining, at a computing device, subject data associated with a subject, the subject data comprising: a) at least one vital parameter determined for the subject, b) at least one laboratory parameter determined for the subject, computing, based on the obtained subject data, at least one first indicator indicative of the appropriateness for multiple overnight hospitalization of the subject.

2. The method of claim 1 , the subject data further comprising: c) information on the subject’s symptoms and / or on the subject’s medical history and / or on interventions undertaken at the subject since hospital admission and / or the subject’s demographic affiliation.

3. The method of claim 1 or 2, the subject data comprising at least one laboratory parameter trend based on the at least one laboratory parameter determined for the subject and measured at a plurality of points in time.

4. The method of any of the preceding claims, the subject data comprising at least one vital parameter trend based on the at least one vital parameter determined for and measured at a plurality of points in time.

5. The method of any of the preceding claims, the method including computing a first indicator trend comprising the at least one first indicator computed for a plurality of points in time, optionally based on the at least one laboratory and / or vital parameter trend of the preceding claims 4 and 3, respectively.

6. The method of any of the preceding claims, further comprising computing, based on the obtained subject data, a time period prediction for a duration of the hospitalization.

7. The method of any of the preceding claims, further comprising computing at least one second indicator indicative of the need for critical care of the subject.

8. The method of any of the preceding claims, further comprising computing at least one third indicator indicative of the need for hospitalization of the subject.

9. The method of any of claims 7 and 8, wherein at least two, optionally three indicators of the at least one the first indicator, the second indicator and / or the third indicator are at least partially computed simultaneously.

10. The method of any of the preceding claims 7 to 9, wherein, based on the at least one first indicator, second indicator and / or third indicator, a composite indicator indicative of the overall health risk of the subject is determined.11 . The method of any of claims 7 to 10, wherein the at least one first indicator determines the categorization of the subject, optionally the composite indicator, if the at least one second indicator indicates no critical care and / or the at least one third indicator indicates hospitalization.

12. The method of any of claims 7 to 11 , wherein the composite indicator indicates the following categories: critical care, multiple overnight hospitalization, short stay hospitalization, and hospital discharge.

13. The method of any of the preceding claims, wherein computing the at least one first indicator, second indicator and / or third indicator is based on at least one machine learning model configured to receive and process input data associated with at least a subset of the subject data.

14. The method of any of the preceding claims, wherein the method includes saving, optionally in an electronic patient file,- at least a subset of the subject data as obtained and / or- at least a subset of the first, second and / or third indicators as computed, and / or- at least the composite indicator as determined.

15. The method of any of the preceding claims, wherein the method includes displaying, at a user interface,- the at least one first, second and / or third indicators of a plurality of subjects, and / or- the at least one first indicator trend determined for a subject at a plurality of points in time, and / or- at least the composite indicator as determined.

16. The method of any of the preceding claims, wherein the at least one laboratory parameter is selected among the group of parameters consisting of:MCHC value;Troponin value;Red blood cell count;White blood cell count;C-reactive protein value;Blood Urea Nitrogen value;Lymphocyte count or value;Hematocrit value;D-dimer value;Hemoglobin value; and RDW value.

17. The method of any of the preceding claims, wherein the at least one vital parameter is selected among the group of parameters consisting of: body temperature; blood pressure; oxygen saturation; heart rate; and respiratory rate.

18. A computing device including one or more processors for data processing, wherein the computing device is configured to carry out the method according to any of the preceding claims 1 to 17.

19. A computer program, which, when executed by one or more processors of a computing device according to claim 18, instructs the computing device to perform the method according to any of the preceding claims 1 to 17.

20. A non-transitory computer-readable medium storing a computer program to carry out the method according to any of the preceding claims 1 to 17.

Images