Systems and methods for early identification of difficult venous access in patient

Abstract

Disclosed herein is a system and method for determining a difficult venous pathway for a patient. The logic processes metadata acquired by a vasculature evaluation device coupled to a patient. The metadata is, for example, vessel diameter, vessel depth, vessel wall thickness, vessel wall elasticity, tissue elasticity, tissue richness, blood flow rate, or hydration level. The logic also determines a difficult venous pathway by executing an algorithm on the metadata. Algorithms are defined using machine learning techniques applied to ongoing collection data sets acquired from multiple systems across multiple patients undergoing catheter insertion events. The dataset may also include patient data, such as weight, age, etc. The vasculature assessment device may include ultrasound imaging, infrared imaging, molecular imaging, Raman spectroscopy, or optical coherence tomography to acquire one or both of three-dimensional imaging data and metadata.

Description

[0001] priority

[0002] This application claims the benefit of priority to U.S. patent application No. 18 / 209,394, filed on June 13, 2023, which is incorporated by reference in its entirety into this application. Technical Field

[0003] The present application relates to the field of medical devices, and more particularly to systems and methods for early identification of difficult venous access in patients. Background Art

[0004] Nurses often face challenges when placing peripheral intravenous (IV) lines in adults and children, a situation described as difficult venous access (DVA). Multiple venipuncture attempts can increase patient anxiety and distress, delay important management, and increase costs. A variety of factors can predispose patients to DVA, including small, fragile, or hidden veins, and veins that collapse due to dehydration are particularly problematic. Multiple placement attempts can exacerbate "needle phobia" because subsequent venipunctures tend to be more painful and are associated with increased rates of extravasation, vessel perforation leading to hematoma or hemorrhage, and phlebitis. Nurses encounter DVA in a variety of clinical settings, including acute medical transport, emergency departments, hospitals, clinicians' offices, long-term care facilities, hospices, and home care. However, there are no assessment scales to predict the degree of difficulty in placing a peripheral IV, nor are there consensus guidelines for preventing or managing these challenges. If patients with potential DVA could be identified early in the process (i.e., before the catheterization event), nurses would have time to adjust their approach and employ special techniques to enhance venous access and improve cannulation success. This in turn reduces the emotional and financial burden associated with repeated failed attempts to place an IV line.

[0005] Disclosed herein are systems and methods for early identification of difficult venous access. Summary of the invention

[0006] According to some embodiments, a medical system is disclosed herein, comprising: a vascular assessment device configured to acquire raw image data of a patient's vascular system; and a system module having a console coupled to the vascular assessment device. The console comprises a processor and a memory having logic stored thereon, the logic performing system operations when executed by the processor, the operations comprising: (i) receiving raw image data from the vascular assessment device; determining metadata of the vascular system from the raw image data; and applying a trained machine learning (ML) model to the metadata to determine a difficult venous access (DVA) assessment for the vascular system.

[0007] In some embodiments, the operations also include determining three-dimensional (3D) imaging data from the raw image data, and applying the trained ML model to the 3D imaging data along with the metadata to determine the DVA assessment.

[0008] In some embodiments, the operations further include determining a map of the vasculature from the 3D imaging data, and depicting the map of the vasculature on a display of the system module.

[0009] In some embodiments, the vascular assessment device acquires raw image data using one or more of ultrasound imaging, infrared imaging, molecular imaging, Raman spectroscopy, or optical coherence tomography.

[0010] In some embodiments, the metadata includes one or more of vessel diameter, vessel depth, blood flow velocity, blood flow direction, vessel wall thickness, vessel wall elasticity, tissue elasticity, tissue profusion, or hydration level.

[0011] In some embodiments, the operations further include providing a DVA assessment report to the clinician, the DVA assessment report comprising one or more of the subset of metadata, a recommended catheter size, a recommended insertion angle, a recommended catheter insertion device, or a DVA score.

[0012] In some embodiments, the suggested catheterization device is selected from a list of catheterization devices stored in memory.

[0013] In some embodiments, the system module is communicatively coupled to an automated or semi-automated catheter insertion device, and the operation includes transmitting the DVA assessment to the automated or semi-automated catheter insertion device for utilization during catheter insertion.

[0014] In some embodiments, the system module is communicatively coupled to an electronic medical record (EMR) system such that the system module exchanges information with the patient's EMR. In some embodiments, exchanging information with the EMR includes transmitting the DVA assessment to the EMR system.

[0015] In some embodiments, the operations further include: (i) receiving patient data for the patient from the EMR system, wherein the patient data includes at least one of the patient's weight, gender, age, race, or body mass index; and (ii) applying the trained ML model to the patient data together with the 3D imaging data and metadata to determine a DVA assessment.

[0016] In some embodiments, the operations also include: (i) identifying multiple locations along the vascular system; (ii) receiving raw image data for each identified location from a vascular assessment device; (iii) determining metadata for each identified location; and (iv) applying the trained ML model to the metadata for each identified location to determine a DVA assessment for each identified location.

[0017] In some embodiments, the system module is communicatively coupled to an external computing device having ML logic stored in a memory, the ML logic performing ML operations when executed by a processor, the ML operations comprising receiving a plurality of historical DVA assessment data sets by the external computing device, wherein each historical DVA assessment data set comprises: a DVA assessment derived from raw image data obtained from a historical patient by one of a plurality of vascular assessment devices; and an independent DVA assessment resulting from actual placement of a catheter in the historical patient. In such embodiments, the ML operations further comprise applying an ML algorithm to the plurality of historical DVA assessment data sets to define a trained ML model, and the logical operations of the system module further comprise receiving the trained ML model from the external computing device.

[0018] In some embodiments, the external computing device is communicatively coupled to the EMR system, and the ML operations further include receiving an independent DVA assessment from the EMR system.

[0019] In some embodiments, the operation of the system module also includes receiving a trained ML model from an external computing device.

[0020] According to some embodiments, a system method for determining a DVA assessment of a patient's vascular system is also disclosed herein, which includes receiving multiple historical DVA assessment data sets, wherein each historical DVA assessment data set includes: metadata and 3D imaging data, which is determined by a vascular assessment device of one of the multiple DVA assessment systems from raw image data acquired by the historical patient; and a corresponding independent DVA assessment, which is generated by actual placement of a catheter in the body of the historical patient, wherein the corresponding independent DVA assessment is recorded in the EMR of the historical patient. The system method also includes: (i) applying an ML algorithm to the multiple historical DVA assessment data sets to train an ML model, the ML model correlating metadata and 3D imaging data with corresponding independent DVA assessments across the multiple historical DVA assessment data sets; (ii) acquiring instant raw image data of the vascular system of the current patient by a vascular assessment device of one of the multiple DVA assessment systems; (iii) determining instant metadata and instant 3D imaging data from the instant raw image data; (iv) applying the trained ML model to the instant metadata and instant 3D imaging data to determine an instant DVA assessment of the vascular system of the current patient, wherein the instant DVA assessment includes a prediction score for the instant metadata and the instant 3D imaging data.

[0021] In some embodiments, the vascular assessment device acquires metadata using one or more of ultrasound imaging, infrared imaging, molecular imaging, Raman spectroscopy, or optical coherence tomography.

[0022] In some embodiments, the metadata includes one or more of vessel diameter, vessel depth, blood flow rate, vessel wall thickness, vessel wall elasticity, tissue elasticity, tissue richness, or hydration level, and in some embodiments, the DVA assessment includes one or more of a recommended catheter size, a recommended insertion angle, a recommended catheter insertion device, or a DVA score.

[0023] In some embodiments, the system method also includes retraining the trained machine learning model using multiple historical DVA assessment data sets and predicted scores for the real-time metadata and real-time 3D imaging data along with independent DVA assessments corresponding to the scores for the real-time metadata and real-time 3D imaging data.

[0024] These and other features of the concepts provided herein will become more readily apparent to those skilled in the art in view of the accompanying drawings and the following description, which describe in more detail specific embodiments of such concepts. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A medical system configured to determine difficult venous access (DVA) of a patient is shown according to some embodiments.

[0026] Figure 2 A portion of vasculature extending along a limb of a patient is shown with a needle inserted into a blood vessel according to some embodiments.

[0027] Figure 3 An exemplary data table is shown of logic executing machine learning techniques used to define a trained ML model that relates difficult venous access (DVA) assessments to metadata about the vasculature, according to some embodiments.

[0028] Figure 4 A block diagram of a systematic method for determining a DVA assessment according to some embodiments is shown.

[0029] FIG. 5A to FIG. 5C Shows Figure 1 An exemplary screenshot of a medical system. DETAILED DESCRIPTION

[0030] Before disclosing some specific embodiments in more detail, it should be understood that the specific embodiments disclosed herein do not limit the scope of the concepts provided herein. It should also be understood that the specific embodiments disclosed herein may have features that can be easily separated from the specific embodiments, and these features may be optionally combined with or replace the features of any of the many other embodiments disclosed herein.

[0031] About the terms used in this article, it should also be understood that these terms are for the purpose of describing some specific embodiments, and these terms do not limit the scope of the concepts provided herein. Ordinal numbers (e.g., first, second, third, etc.) are generally used to distinguish or identify different features or steps in a set of features or steps, and do not provide sequence or numerical restrictions. For example, "first", "second" and "third" features or steps do not necessarily appear in this order, and the specific embodiments including such features or steps are not necessarily limited to these three features or steps. For convenience, labels such as "left", "right", "top", "bottom", "front", "back", etc. are used, and these labels are not intended to imply, for example, any specific fixed position, orientation or direction. On the contrary, such labels are used to reflect, for example, relative position, orientation or direction. The singular forms of "a kind of", "one" and "the" include plural references unless the context clearly stipulates otherwise. For example, "vascular parameter" used herein can be a single vascular parameter or a combination of multiple single vascular parameters.

[0032] The phrases "connected to," "coupled to," and "in communication with" refer to any form of interaction between two or more entities, including but not limited to mechanical, electrical, magnetic, electromagnetic, fluid, wireless, and thermal interactions. Two components may be coupled to each other even if they are not in direct contact with each other. For example, two components may be coupled to each other through an intermediate component.

[0033] The term "logic" may refer to hardware, firmware, or software configured to perform one or more functions. As hardware, the term "logic" may refer to or include circuits having data processing and / or storage functions. Examples of such circuits may include, but are not limited to or constrained to, hardware processors (e.g., microprocessors, one or more processor cores, digital signal processors, programmable gate arrays, microcontrollers, application specific integrated circuits "ASICs", etc.), semiconductor memories, or combinational components.

[0034] Additionally or alternatively, the term "logic" may refer to or include software, such as one or more processes, one or more instances, an application programming interface (API), a subroutine, a function, an applet, a servlet, a routine, a source code, an object code, a shared library / dynamic link library (dll), or even one or more instructions. The software may be stored in any type of suitable non-transitory storage medium or transient storage medium (e.g., an electrical, optical, acoustic, or other form of propagating signal, such as a carrier wave, an infrared signal, or a digital signal). Examples of non-transitory storage media may include, but are not limited to or limited to: programmable circuits; non-persistent storage devices, such as volatile memory (e.g., any type of random access memory "RAM"); or persistent memory, such as non-volatile memory (e.g., read-only memory "ROM", power-powered RAM, flash memory, phase change memory, etc.), a solid-state drive, a hard drive, an optical drive, or a portable memory device. As firmware, the logic may be stored in a persistent storage device.

[0035] Unless defined otherwise, all technical and scientific terms used herein have the same meanings as commonly understood by one of ordinary skill in the art.

[0036] Any method disclosed herein includes one or more steps or actions for performing the described method. These method steps and / or actions can be interchangeable with each other. In other words, unless the proper operation of the embodiment requires a specific order of steps or actions, the order and / or use of specific steps and / or actions can be modified. In addition, a subroutine or only a portion of the method described herein can be a separate method within the scope of the disclosed text. In other words, some methods can include only a portion of the steps described in the more detailed method. In addition, all embodiments disclosed herein are combinable and / or interchangeable, unless otherwise stated, or such combination or interchange will be contrary to the operability of any embodiment.

[0037] Figure 1 A medical system 100 is shown that is configured to determine (i.e., identify or diagnose) a patient's difficult venous access according to some embodiments. In some embodiments, the system 100 can determine that access to the vein may be difficult or require special tools, equipment, or procedures to successfully place a catheter (or other medical device) in the vein, and can provide notifications accordingly. The medical system 100 typically includes a plurality of difficult venous access (DVA) assessment systems 110 that are communicatively connected to an external computing device 60. The external computing device 60 can be, for example, any suitable computing device, such as a network server. In some embodiments, the external computing device 60 can be connected to an electronic medical record (EMR) system. Machine learning (ML) logic 109 (including machine learning algorithm 109A) and database 107 are stored in a memory (e.g., a non-transitory computer-readable medium) of the external computing device 60.

[0038] DVA assessment system (system) 110A is one of the DVA assessment systems 110, which is shown for use with a patient 50 undergoing a DVA assessment procedure. System 110A is generally configured to obtain a map of a vasculature 55 of patient 50 and to acquire metadata related to the vasculature 55. System 110A (like each of the systems 110) includes a system module 111 coupled to a vasculature assessment device (device) 150. Device 150 is configured to acquire raw image data and provide the raw image data to system module 111. Logic of system module 111 may then determine three-dimensional (3D) imaging data of vasculature 55 from the raw image data (a map of vasculature 55 may be determined from the 3D imaging data). Logic of system module 111 is also configured to determine metadata related to vasculature 55 from the raw image data. The map of vasculature 55 may include identification of a plurality of blood vessels available for access. The metadata may include one or more of vessel diameter, vessel depth, blood flow rate, vessel wall thickness, vessel wall elasticity, tissue elasticity, tissue richness, or hydration level.

[0039] The device 150 can be operatively coupled to the patient 50 in any suitable manner. In some embodiments, the device 150 can surround or partially surround a limb of the patient 50, wherein the vasculature 55 extends along the limb. In some embodiments, the device 150 can, for example, include operative physical contact with the patient, such as via a probe. The device 150 can include one or more of any suitable technologies configured to obtain raw image data. In some embodiments, suitable technologies can include ultrasound imaging, infrared imaging, molecular imaging, Raman spectroscopy, or optical coherence tomography to acquire raw image data.

[0040] The system 110A includes a console 115, and the device 150 is communicatively coupled to the console 115, such as via a wired connection or a wireless connection, for example. In the illustrated embodiment, the console 115 includes a processor (or multiple processors) 116 that executes logic (such as imaging logic 118 and DVA logic 119) stored in a memory 117 (e.g., a non-transitory computer-readable medium). The power supply 114 of the console 115 may include an external power source (i.e., a facility power source) or a battery. The console 115 may include a wireless module 113 to facilitate wireless communication between the device 150 and the system module 110A and / or between the system module 110A and an external computing device 60. In some embodiments, the system 110A may include (or be coupled to) a display 112 (e.g., a graphical user interface) that is configured to display notification information and / or receive input from a clinician.

[0041] In some embodiments, imaging logic 118 is configured to receive raw image data from device 150 and depict an image (or map) 125 of vasculature 55 on display 112. Thus, image 125 may enable a clinician to identify a target vessel (e.g., a target vein) or to select a different target vessel. Image 125 may enable a clinician to identify an insertion site along a target vessel. Similarly, image 125 may, for example, enable a clinician to identify any abnormalities associated with vasculature 55, such as stenosis or blockage of a vessel. In some embodiments, image 125 may, for example, depict vasculature 55 relative to other anatomical elements, such as bones, nerves, or other blood vessels. In FIG. 5A to FIG. 5C An example of an image (map) 125 is shown in FIG.

[0042] The DVA logic 119 is generally configured to receive raw image data from the device 150 and determine metadata from the raw image data. The DVA logic 119 may also be configured to determine a difficult venous access (DVA) assessment for the vasculature 55. The metadata may include data related to accessing the vasculature 55. Thus, the metadata may include properties of the vasculature 55 (or more specifically, a target vessel) and properties of tissue or other anatomical elements proximate to the vasculature 55.

[0043] In some embodiments, the metadata may additionally include patient data, which in some embodiments may be entered by a clinician or collected from an EMR of the patient 50. The patient data may include, for example, any suitable patient parameters, such as height, weight, gender, age, race, limb circumference, or body mass index (BMI). Other patient conditions may also be included, such as blood pressure, pulse rate, medications, or body temperature.

[0044] DVA assessment can include any information that helps clinicians determine one or more suitable (or best) programs, instruments or equipment for entering vascular system 55. In some embodiments, DVA assessment can include target blood vessel, insertion site, and / or pin or limit the insertion angle of catheter. In some embodiments, DVA assessment can include advising catheter type or size. In some embodiments, DVA assessment can include advising catheter insertion instrument or system (for example, automation or semi-automatic catheter insertion system). DVA assessment can for example include difficulty classification, such as simple, medium, difficult or extremely difficult or numerical score.

[0045] The DVA logic 119 can provide a notification or report to the clinician. For example, the DVA logic 119 can provide a visual report on the display 112. The DVA logic 119 can also provide an alert or warning in the form of a visual or audible indication based on the DVA assessment. In some embodiments, the DVA logic 119 can transmit the DVA assessment to the EMR of the patient 50. Therefore, the DVA assessment can enable the clinician to perform a subsequent catheterization for the patient 50.

[0046] In the illustrated embodiment, the DVA logic 119 determines the DVA assessment by utilizing machine learning techniques, for example, applying a trained machine learning model 109A to specific inputs, which may include all or any subset of the 3D imaging data and / or metadata. In some embodiments, the trained machine learning model 109A may be stored in the memory 117. In other embodiments, the trained machine learning model 109A may be stored in a database 107 of an external computing device 60 and executed by the external computing device 60. In such embodiments, the DVA logic 119 may transmit the 3D imaging data and / or metadata to the external computing device 60 and receive a DVA assessment from the external computing device 60, wherein the DVA assessment may include a point or location along the patient's vasculature that is predicted to be the easiest to enter, as determined by processing the 3D imaging data and / or metadata by the trained machine learning model 109A.

[0047] The ML logic 109, when executed by one or more processors, is typically configured to process historical data, thereby generating a trained machine learning model 109A. The ML logic 109 receives historical data from multiple catheter placement events across a group of patients as input. In some embodiments, the external computing device 60 can be communicatively connected to the EMR system and / or multiple systems 110 and configured to receive historical data therefrom. The historical data for each catheter placement event typically includes 3D imaging data, metadata, and final catheter placement data. The historical data used to train the machine learning model 109A can include a collection of the following items: (1) 3D imaging data; (2) metadata; and (3) an independent DVA assessment, i.e., a DVA assessment generated by the actual placement of the catheter. In some cases, only one of the 3D imaging data and the metadata is utilized. As a detailed example, a series of points along the patient's vascular system and metadata for each point can be identified. Metadata for a particular point may include vessel diameter, vessel depth, blood flow rate, vessel wall thickness, vessel wall elasticity, tissue elasticity, tissue richness, or hydration level, and may typically be associated with patient data for the patient, such as any of the patient's weight, gender, age, race, or body mass index.

[0048] The independent DVA assessment may include any relevant data or information related to the actual insertion of the catheter. For example, the relevant data or information may include the catheter size used. According to one embodiment, a list of available catheters (or catheter sizes) may be stored in the memory 117. Thus, the DVA assessment may include a suggested catheter size (or model) selected from the list by the DVA logic 119 as a result of applying the trained ML model 109A. In a similar manner, the DVA assessment may include a suggested catheter insertion device and / or insertion angle.

[0049] In some embodiments, the catheter insertion device can be automated or semi-automatic. In such embodiments, the catheter insertion device (or more specifically, its operation) can be configured to utilize metadata, 3D imaging data, and / or DVA assessment to limit the setting of the catheter insertion device or otherwise affect the operation of the catheter insertion device. Therefore, according to an embodiment, system 100 can be configured to be connected with the catheter insertion device and provide data thereto. As an example, the relevant data or information relevant to the actual insertion of conduit can be the operation settings of the catheter insertion device, and the DVA assessment can include these operation settings.

[0050] Raw image data may be obtained by deploying any of the following medical systems: an ultrasound imaging system, an infrared imaging system, a molecular imaging system, a Raman spectroscopy system, and / or an optical coherence tomography system. For example, an ultrasound system may include an ultrasound probe configured to project high-frequency sound waves through body tissue and receive echoing sound waves to create raw image data about the patient's vasculature and / or other anatomical elements. The logic executed by the processor may then create an image (i.e., a map) 125 of the vascular system 55 suitable for presentation on the display 112, wherein the image 125 may, for example, include properties of the vascular system, such as location or depth from the skin surface, vessel wall thickness, vessel size, or vessel shape. The ultrasound system may also utilize Doppler technology to detect / determine the movement of anatomical elements, such as blood flow rate or vascular pulsation. Monitoring and / or analyzing vascular pulsation (i.e., stretching and contraction due to pressure within it) may be used to indicate the elasticity or hardness of the vessel. Other systems (e.g., imaging systems) may, for example, utilize similar techniques to obtain raw data related to other properties of the patient's vascular system (such as tissue richness or hydration level).

[0051] Thus, ML logic 109 may apply an ML algorithm to a collection of 3D imaging data, metadata, and independent DVA assessments (which serve as a label as to whether the clinician successfully entered the vasculature at the corresponding point) to generate a trained ML model 109A, which may include a series of weighted features that, when applied to the input of the 3D imaging data and / or metadata, determine a score associated with a predicted likelihood that vasculature entry at the corresponding point will be successful in view of the historical data. It should be appreciated that the historical data may be stored in memory 117. Although Figure 1 The database 107 and ML logic 109 are shown as being stored on the external computing device 60 , but in some embodiments, the database 107 and ML logic 109 may be stored in the memory 117 of the console 115 , such that the ML logic 109 can be executed by the processor 116 .

[0052] Figure 2 A portion of a patient's limb 250 is shown having vasculature 256 extending along it. A needle 260 is shown inserted through skin surface 253 at insertion site 257 and further inserted into a target vessel 256 at insertion site 257. As shown, vasculature 255 includes more than one vessel that can be accessed via needle 260. In some embodiments, the DVA assessment can include indicating that one of the multiple vessels can be most easily accessed, i.e., the DVA assessment can indicate an optimal vessel, e.g., target vessel 256, based on at least one of 3D imaging data or metadata. In some embodiments, the DVA assessment can also indicate an optimal insertion angle 261 for needle 260.

[0053] Figure 3 An exemplary table 300 of a database 107 according to some embodiments is shown. The table 300 includes multiple (e.g., 10, 100, 1000, or more) data sets (or records) 310 received from multiple systems 110, wherein each data set 310 may include data related to a single vascular access / catheterization event for a specific patient. The data table 300 may include metadata 320, patient data 330, and / or vascular access data 340. In some embodiments, the patient data 330 may be omitted.

[0054] Metadata 320 may include metadata collected by device 150 as described above, such as vessel diameter, vessel depth, blood flow rate, vessel wall thickness, vessel wall elasticity, tissue elasticity, tissue richness, or hydration level, as shown in table 300. Of course, metadata 320 may not be limited to the metadata shown, and in some embodiments, metadata 320 may include only a subset of the metadata shown.

[0055] Patient data 330 may include the weight, gender, age, race, body mass index (BMI) of a particular patient. Of course, patient data 330 may not be limited to the patient data shown. For example, additional patient data (not shown) may include blood pressure, medications, diseases, or any other medical condition of the patient. In some embodiments, patient data 330 may include only a subset of the patient data shown. In some embodiments, patient data 330 may be recorded in an EMR for a particular patient, and thus, external computing device 60 may receive patient data 330 from the EMR system.

[0056] Vascular access data 340 are data related to the result of vascular access / catheterization event, which can constitute an independent DVA assessment, that is, a DVA assessment produced by the actual placement of a catheter. Vascular access data 340 can include the blood vessel entered, the catheter used (e.g., type and / or size), any catheter insertion equipment used, insertion site, insertion angle, and DVA score or grading, as shown in table 300. Of course, in other embodiments, vascular access data 340 can include more or less vascular access data than the vascular access data shown. In some cases, vascular access data 340 can be recorded in the EMR of a specific patient. Therefore, in some embodiments, external computing device 60 can receive vascular access data 340 from an EMR system.

[0057] The ML logic 109 is generally configured to apply an ML algorithm to the historical data to define a relationship between the metadata 320 (and in some embodiments, the patient data 330) and the vascular access data 340. The ML logic 109 can then define a trained ML model 109A based on the relationship, wherein the trained ML model 109A is configured to indicate difficulty in accessing the vascular system of the instant patient based on the metadata collected from the instant patient. The DVA logic 119 is configured to collect the instant metadata and apply the trained ML model to the instant metadata to determine an instant DVA assessment for the instant patient 50.

[0058] Figure 4 is a block diagram of a system method for determining a DVA assessment of a patient's vasculature, according to some embodiments, the system method including all or any subset of the following actions, steps, or processes. Figure 4Each box shown in represents an operation of the method 400 performed by the medical system disclosed herein, and is generally a result of executing one or more logic modules disclosed herein and deploying specific machines or devices (such as vascular system assessment device 150, ultrasound probe, etc.). The method 400 may include obtaining multiple historical catheter placement data sets (box 410), wherein each data set includes 3D imaging data and metadata related to the patient's vascular system (collectively referred to as "data sets"), wherein each data set originates from one or more medical systems in the multiple medical systems, and wherein each medical system can acquire a separate "data set". Examples of medical systems are discussed above and include, but are not limited to, or are constrained to, ultrasound imaging systems, infrared imaging systems, molecular imaging systems, Raman spectroscopy systems, and / or optical coherence tomography systems. Each medical system is configured to acquire a patient's data set, for example, before the patient experiences a catheter placement event, and each data set is associated with, for example, a final catheter insertion data set recorded by a clinician (sometimes referred to herein as vascular access data), wherein the final catheter placement data set defines an independent DVA assessment. Therefore, the medical system obtains a data set (3D imaging data and metadata) and then associates it with the final catheter placement data set. The collection of data sets and associated final catheter placement data sets relative to box 410 can be considered historical data, which can be used to train the machine learning model discussed below or used when analyzing future (non-historical) data sets by performing certain heuristics. In some embodiments of method 400, each medical system includes: (i) a vascular assessment device that acquires raw image data using one or more of ultrasound imaging, infrared imaging, molecular imaging, Raman spectroscopy, or optical coherence tomography; and (ii) logic that, when executed by a processor, determines medical data from the raw image data, the medical data can include one or more of 3D imaging data and / or metadata.

[0059] Method 400 may also include executing logic stored on a non-transitory computer-readable medium by one or more processors, resulting in executing a machine learning algorithm with historical data as input, thereby generating a trained machine learning model configured to provide a prediction of accessibility difficulty of a point along the patient's vascular system (box 420). For example, the processing of the historical data by the machine learning algorithm produces an automated weighting of various features (e.g., aspects of the 3D imaging data and metadata) in view of the catheter placement data, so that future execution of the trained machine learning model with current 3D imaging data and / or metadata as input will produce a score reflecting the difficulty of entering the vascular system at a point along the vascular system corresponding to the 3D imaging data and metadata (e.g., based on the historical data indicating whether the clinician will obtain vascular system access at that point). In some embodiments of method 400, the metadata may include one or more of vascular diameter, vascular depth, blood flow rate, vascular wall thickness, vascular wall elasticity, tissue elasticity, tissue richness, or hydration level. The trained machine learning model may be stored in the memory 117 of the console.

[0060] Method 400 may also include acquiring instant raw image data and / or metadata for the current patient via deployment of a vasculature assessment device (e.g., device 150) as a source of one or more data sets of the current patient's vasculature (box 430), wherein each data set includes 3D imaging data and / or metadata related to the current patient's vasculature and originates from one or more DVA systems including the vasculature assessment device or integrated with the vasculature assessment device in a plurality of DVA systems. After deployment of the vasculature assessment device results in acquisition of raw image data and conversion from raw image data to 3D imaging data and metadata, method 400 may also include performing a logical application to determine a DVA assessment for the current patient (box 440). Embodiments of determining a DVA assessment for the current patient are discussed herein and include the use of heuristics and / or machine learning techniques. For example, a trained ML model may be applied to the instant 3D imaging data and / or metadata to generate a score for a given point along the patient's vasculature, the score reflecting the difficulty of entering the vasculature at that point. Brief Reference Figure 5A , a plurality of points (circles) are shown along the vasculature such that a trained machine learning model can determine a score for each point. For example, in some embodiments, a trained ML model can be configured to evaluate points along the vasculature at predetermined intervals (e.g., every three (3) inches along the length of the imaged vasculature or imaged portion of the vasculature for which metadata is determined). In some embodiments of method 400, the DVA evaluation includes one or more of an insertion site or an insertion angle.

[0061] In some embodiments, method 400 also includes retraining the trained machine learning model using the historical data and the predicted scores for the current 3D imaging data and metadata, along with an independent DVA assessment corresponding to the score for the current 3D imaging data and metadata (e.g., an indication of which point along the vasculature was entered and / or whether an entry failure occurred at a particular point) (block 450). In some embodiments, the historical data will be assigned a first weight, and the current 3D imaging data and metadata (with the independent DVA assessment as a label) will be assigned a second weight that is greater than the first weight. In some embodiments, no weights are assigned (i.e., all data is weighted equally). The retrained machine learning model can then be used to analyze future 3D imaging data and metadata.

[0062] Figure 5A An exemplary screenshot 510 is shown, which may be depicted on display 112 as an output of graphics logic 118 and / or DVA logic 119. Screenshot 510 includes image 511 (e.g., Figure 1 512 ). The image includes a map of a vasculature 55 including a blood vessel 512. Non-vascular anatomical elements 513 (e.g., bones) are also shown in relation to the blood vessels 512. Thus, the clinician can visualize the blood vessels 512 in relation to the bones, which the clinician can use as input when making vascular access decisions. Discrete locations along the blood vessels 512 are also shown, as indicated by circles, where these discrete locations may be possible entry points (insertion sites) for the blood vessels 512.

[0063] Also shown in the screenshot 510 is a DVA report 520. The DVA report 520 includes metadata, such as vessel depth and vessel diameter. The DVA report 520 also includes various DVA outputs of the trained ML model when applied to the metadata. In the example shown, the DVA output includes a suggested catheter size (e.g., 5 French) and a suggested insertion angle (e.g., 15 degrees). The DVA output also includes a suggestion that no insertion tools are required to successfully enter the blood vessel 512.

[0064] In some cases, a vessel may be relatively constant in size and accessibility, such as vessel 512, for example, over a defined length. As described above, DVA logic 119 (or more specifically, a trained ML model) may define a DVA score (i.e., a rating associated with difficulty in accessing the vessel). According to one example, the score may be numeric, e.g., 1 to 10, with a score of 1 indicating easy or simple access and a score of 10 indicating difficult or complex access. In this example, DVA logic 119 has determined a DVA score for each discrete location along vessel 512, and the DVA score for each location is depicted in image 511 adjacent to each respective discrete location. In some embodiments, DVA logic 119 may determine a composite DVA score for vessel 512, and the composite DVA score may be included in report 520.

[0065] Figure 5B Another exemplary screenshot 530 is shown, which can be depicted on the display 112 as an output of the image logic 118 and / or the DVA logic 119. The screenshot 530 includes an image 531, which includes a mapping of the vascular system 55, which includes blood vessels 532, 533, and 534. The mapping of the vascular system can enable the clinician to visually observe any abnormalities of the vascular system, such as the stenosis 534A of the blood vessel 534. Discrete locations along the blood vessels 532, 533, and 534 as indicated by circles are also shown. The DVA report 540 includes metadata, such as vascular depth and vascular diameter. The DVA report 540 includes a suggested catheter size (e.g., 4 French) and a suggested insertion angle (e.g., 30 degrees). The DVA output also includes a suggestion that a needle guide can be beneficial for successfully entering the vascular system 55.

[0066] In the example shown, various DVA outputs are depicted in image 531 along with a map of the vascular system 55. As shown, some metadata are indicated as being adjacent to the corresponding discrete locations, such as DVA scores, blood flow rates, and vessel diameters. In some embodiments, imaging logic 118 and / or DVA logic 119 can enable a clinician to determine via input which metadata and / or DVA outputs will be depicted in image 531 and / or a DVA report, as shown. In other words, imaging logic 118 and / or DVA logic 119 can include filters that allow clinicians to turn on / off the visibility of certain information.

[0067] In addition to or in lieu of depicting a DVA score #, the circle (or dot) may, for example, include other visual indication elements such as color, shape, border, or X-out. For example, a dot including an "X" may indicate a location that is not recommended for entry, and a dot including a border 536 may indicate a location that is recommended for entry. Although not shown, the border 536 may also include a specific shape to indicate a preference. For example, a square border may indicate a recommended first entry choice, a pentagonal border may indicate a recommended second entry choice, and a hexagonal border may indicate a recommended third entry choice.

[0068] Other options not shown may include a binary indication method, including a black dot indicating a location that is not recommended for entry and a white dot indicating a location that is recommended for entry. Similarly, a variable indication method may include a color spectrum (e.g., white-gray-black or green-yellow-red) based on the DVA score. For example, a white dot may indicate a DVA score below "4", a gray dot may indicate a DVA score of "4 to 7", and a black dot may indicate a DVA score above "7". Although the above examples utilize color or other visual indicators based on the DVA score, color or any other markings may be used, for example, to indicate specific aspects of the DVA assessment, such as blood flow rate, vessel depth, and / or vessel diameter.

[0069] Although Figure 5A The DVA report 520 of FIG. 540 relates to the vessel 512 as a whole (e.g., the DVA score is a composite score), but the DVA report 540 can be related to a single discrete location, such as the location including the bounding box 536 or any location that can be selected by the clinician via input. In other words, the system can enable the clinician to view the DVA report for any discrete location by selecting the location.

[0070] Figure 5C Another exemplary screenshot 550 is shown that may be depicted on the display 112 as an output of the image logic 118 and / or the DVA logic 119. Instead of indicating discrete locations, the image 551 shows a vessel 552 having a DVA score variably (e.g., continuously) indicated along the length of the vessel 552 depicted in the image 551. In other words, the vessel 552 (i.e., the image of the vessel 552) itself includes a color spectrum associated with the DVA score. As an example, as shown, as the DVA score changes from 1 to 10 along the depicted length, the color of the vessel image gradually transitions from white to gray to black. Similar to Figure 5BAs an example, the color of the vessel image can be associated with a DVA score or a score or metadata for any specific aspect of the DVA assessment, which can be selected by the clinician via input. As another example (not shown), the vessel image can include a plurality of discrete stripes extending along the length of the vessel image, wherein each stripe includes a color spectrum according to a score or metadata for any specific aspect of the DVA assessment. The DVA report 560 includes metadata and / or DVA aspects or outputs associated with a selected portion or location along the vessel image.

[0071] Although some specific embodiments have been disclosed herein, and although these specific embodiments have been disclosed in considerable detail, these specific embodiments are not intended to limit the scope of the concepts provided herein. Additional adaptations and / or modifications may be envisioned by those of ordinary skill in the art, and in broader aspects, these adaptations and / or modifications are also encompassed. Therefore, changes may be made to the specific embodiments disclosed herein without departing from the scope of the concepts provided herein.

Claims

1. A medical system for identifying difficult venous access in a patient, comprising: a vascular assessment device configured to acquire raw image data of the patient's vascular system; and A system module having a console coupled to the vascular assessment device, the console comprising a processor and a memory having logic stored thereon, the logic when executed by the processor performing operations comprising: receiving the raw image data from the vascular assessment device; determining metadata of the vasculature from the raw image data; and A trained machine learning model is applied to the metadata to determine a difficult venous access assessment for the vasculature.

2. The medical system of claim 1, wherein the operations further comprise: determining three-dimensional imaging data from the raw image data; as well as The trained machine learning model is applied to the three-dimensional imaging data along with the metadata to determine the difficult venous access assessment.

3. The medical system of claim 2, wherein the operations further comprise: determining a map of the vasculature from the three-dimensional imaging data; as well as The map of the vascular system is depicted on a display of the system module.

4. The medical system of any one of the preceding claims, wherein the vascular assessment device acquires the raw image data using one or more of ultrasound imaging, infrared imaging, molecular imaging, Raman spectroscopy, or optical coherence tomography.

5. A medical system according to any of the preceding claims, wherein the metadata includes one or more of vessel diameter, vessel depth, blood flow velocity, blood flow direction, vessel wall thickness, vessel wall elasticity, tissue elasticity, tissue richness or hydration level.

6. The medical system of any of the preceding claims, wherein the operations further comprise providing a difficult venous access assessment report to a clinician, the difficult venous access assessment report comprising one or more of a subset of the metadata, a recommended catheter size, a recommended insertion angle, a recommended catheter insertion device, or a difficult venous access score.

7. The medical system of claim 6, wherein the suggested catheterization device is selected from a list of catheterization devices stored in the memory.

8. The medical system of claim 6, wherein: The system module is communicatively coupled to an automated or semi-automated catheter insertion device, and The operation includes transmitting the difficult venous access assessment to the automated or semi-automated catheterization device.

9. The medical system of any of the preceding claims, wherein the system module is communicatively coupled to an electronic medical record system such that the system module exchanges information with the patient's electronic medical record.

10. The medical system of claim 9, wherein exchanging information with the electronic medical record comprises transmitting the difficult venous access assessment to the electronic medical record system.

11. The medical system of claim 9, wherein the operations further comprise: receiving patient data of the patient from the electronic medical record system, the patient data comprising at least one of weight, gender, age, race, or body mass index of the patient; as well as The trained machine learning model is applied to the patient data along with the three-dimensional imaging data and the metadata to determine the difficult venous access assessment.

12. The medical system of any one of the preceding claims, wherein the operations further comprise: identifying a plurality of locations along the vasculature; receiving, for each identified location, the raw image data from the vascular assessment device; determining metadata for each identified location; as well as The trained machine learning model is applied to the metadata for each identified location to determine a difficult venous access assessment for each identified location.

13. The medical system of any of the preceding claims, wherein the system module is communicatively coupled to an external computing device having machine learning logic stored in a memory, the machine learning logic when executed by a processor performing machine learning operations comprising: A plurality of historical difficult venous access assessment data sets are received by the external computing device, each historical difficult venous access assessment data set comprising: a difficult venous access assessment derived from raw image data obtained from a historical patient by one of a plurality of vascular assessment devices; and Independent difficult venous access assessment resulting from actual placement of a catheter in the patient with the history; and applying a machine learning algorithm to the plurality of historical difficult venous access assessment datasets to define the trained machine learning model, and The operations also include receiving the trained machine learning model from the external computing device.

14. The medical system of claim 13, wherein the external computing device is communicatively coupled to the electronic medical record system, the machine learning operation further comprising receiving the independent difficult venous access assessment from the electronic medical record system.

15. The medical system of claim 13, wherein the operations further comprise receiving the trained machine learning model from the external computing device.

16. A system method for determining a difficult venous access assessment of a patient's vasculature, comprising: Receiving a plurality of historical difficult venous access assessment data sets, each historical difficult venous access assessment data set comprising: metadata and three-dimensional imaging data determined from raw image data acquired from a historical patient by a vascular assessment device of one of the plurality of difficult venous access assessment systems; and a corresponding independent difficult venous access assessment resulting from actual placement of a catheter in said historical patient, said corresponding independent difficult venous access assessment being documented in said historical patient's electronic medical record; applying a machine learning algorithm to the plurality of historical difficult venous access assessment datasets to train a machine learning model that correlates the metadata and three-dimensional imaging data with the corresponding independent difficult venous access assessments across the plurality of historical difficult venous access assessment datasets; The vascular assessment device of one of the multiple difficult venous access assessment systems acquires instant raw image data of the vascular system of the current patient; determining instantaneous metadata and instantaneous three-dimensional imaging data from the instantaneous raw image data; and The trained machine learning model is applied to the instantaneous metadata and the instantaneous three-dimensional imaging data to determine an instantaneous difficult venous access assessment of the vasculature of the current patient, the instantaneous difficult venous access assessment comprising a prediction score for the instantaneous metadata and the instantaneous three-dimensional imaging data.

17. The system method of claim 16, wherein the vascular assessment device acquires the raw image data using one or more of ultrasound imaging, infrared imaging, molecular imaging, Raman spectroscopy, or optical coherence tomography.

18. The system method of claim 16 or 17, wherein the metadata comprises one or more of vessel diameter, vessel depth, blood flow rate, vessel wall thickness, vessel wall elasticity, tissue elasticity, tissue richness, or hydration level.

19. The system method of any one of claims 16 to 18, wherein the difficult venous access assessment comprises one or more of a suggested catheter size, a suggested insertion angle, a suggested catheter insertion device, or a difficult venous access score.

20. The system method of any one of claims 16 to 19, further comprising retraining the trained machine learning model using the plurality of historical difficult venous access assessment data sets and the predicted scores of the instant metadata and the instant three-dimensional imaging data and together with independent difficult venous access assessments corresponding to the scores of the instant metadata and the instant three-dimensional imaging data.

Images