WO2021247557A1

WO2021247557A1 - Artificial intelligence and/or machine learning based systems, devices, and methods for designing patient prescriptions

Info

Publication number: WO2021247557A1
Application number: PCT/US2021/035228
Authority: WO
Inventors: Farrukh USMAN; Fouad BAJWA; Sarim JALIL; Ali NUAMAN; Rasib MOOSVI
Original assignee: Byonyks Medical Devices, Inc.
Priority date: 2020-06-01
Filing date: 2021-06-01
Publication date: 2021-12-09

Abstract

Methods and systems for generating a prescription using an artificial intelligence or machine learning model are described herein. The method can include receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and one or more data items associated with a treatment of the patient, and using the machine learning model to determine a recommended prescription for the patient based on the one or more data items associated with the patient and one or more data items associated with the treatment of the patient. The method can also include transmitting the recommended prescription to a medical professional for approval, and in response to receiving user input from the medical professional, generating a final prescription based on the received user input, and storing the final prescription for later transmittal to the therapeutic machine.

Description

ARTIFICIAL INTELLIGENCE AND/OR MACHINE LEARNING BASED SYSTEMS, DEVICES, AND METHODS FOR DESIGNING PATIENT

PRESCRIPTIONS

CROSS-REFERENCE TO RELATED APPLICATION(S)

[0001] This application claims the benefit of U.S. Provisional Patent Application No. 63/032,751, filed June 1, 2020, which is incorporated by reference herein in its entirety.

TECHNICAL FIELD

[0002] The present disclosure is directed to artificial intelligence and/or machine learning based systems, devices, and methods. For example, implementations of the present technology are directed to an Internet of Things (“IoT”) based artificial intelligence and/or machine learning platform that evaluates a medical record and/or other information of a patient with kidney failure and accordingly designs a prescription for the patient.

BACKGROUND

[0003] Peritoneal dialysis (“PD”) is a medical procedure administered when the kidneys perform insufficient functions. The treatment/procedure removes toxins from the human body through their peritoneal membrane. These toxins take advantage of the semi-permeable membrane which surrounds the walls of the peritoneal cavity. During a PD procedure, dialysate solution is introduced into a patient’ s abdomen and remains there for several hours to remove toxins via osmotic transfer. The solution and toxins are then drained from the patient’ s abdomen. There are two forms of PD procedures: continuous ambulatory peritoneal dialysis (CAPD) and automated peritoneal dialysis (APD). During CAPD, patients manually drain the solution. A medical device, such as an APD machine or pump, performs APD procedures.

[0004] Chronic Kidney Disease (CKD) is classified into five stages that indicate the condition and function-quality of a kidney. Stage 1 is used to refer to kidneys with a normal or high glomerular filtration rate (GFR) of 90 mL/min or greater; Stage 2 is considered mild CKD and is used to refer to kidneys with a GFR between 60 mL/min and 89 mL/min; Stage 3A is considered moderate CKD and is used to refer to kidneys with a GFR between 45 mL/min and 59 mL/min; Stage 3B is also considered moderate CKD and is used to refer to kidneys with a GFR between 30 mL/min and 44 mL/min; Stage 4 is considered severe CKD and is used to refer to kidneys with a GFR between 15 ml 7m in and 29 mL/min; and Stage 5 is considered end stage CKD and is used to refer to kidneys with a GFR less than 15 mL/min. Patients with kidney failure are those diagnosed with stage 4 or stage 5 CKD. At stages 4 and 5, a patient’s kidney is moderately or severely damaged and is unable to function correctly. Therefore, the patient’s blood pressure and red blood cells must be closely monitored and managed to ensure any built- up waste in the patient’s body does not create complications.

[0005] Kidney treatments and procedures, including PD, often require a professionally skilled person’ s attention for better patient outcomes. Thus, nephrologists routinely prescribe PD prescriptions to their patients and actively monitor the patient’s health progress and records to determine whether any updates or amendments to a patient’ s PD treatments are warranted. This is becoming an increasingly difficult task, however, as the number of kidney failure patients continues to increase while the availability of skilled professionals decreases.

BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Many aspects of the present disclosure can be better understood with reference to the following drawings. The components in the drawings are not necessarily to scale. Instead, emphasis is placed on illustrating clearly the principles of the present disclosure. The drawings should not be taken to limit the disclosure to the specific implementations depicted, but are for explanation and understanding only.

[0007] FIG. 1 is a partially schematic block diagram illustrating components of a prescription software system configured in accordance with various implementations of the present technology.

[0008] FIG. 2 is a flow chart illustrating a method for generating a prescription using a machine learning model in accordance with various implementations of the present technology.

[0009] FIG. 3 is a partially schematic block diagram of a system including a software cloud configured in accordance with various implementations of the present technology.

DETAILED DESCRIPTION

[0010] The present disclosure is directed to systems, devices, and methods for designing prescriptions for patients using an artificial intelligence and/or machine learning model. In some implementations, the prescription is designed based on the patient’s medical history; PD cycler or other medical device; smart, wearable, and/or web-enabled device; doctor input; happiness index value (generated from the patient’s social media, financial investments, local news, global news, etc.), genealogical data, and/or other data. In the illustrated implementations below, a prescription software service of the present technology is primarily described in the context of designing prescriptions for peritoneal dialysis treatments. Prescription software services in accordance with various implementations of the present technology, however, can be incorporated into and/or used by other systems, including hemodialysis systems and/or other medical or non-medical systems. Furthermore, a person skilled in the art will understand (i) that the technology may have additional implementations than illustrated in FIGS. 1-3 and (ii) that the technology may be practiced without several of the details of the implementations described below with reference to FIGS. 1-3.

A. Overview

[0011] As discussed above, many medical conditions and therapies often require a doctor’s or medical professional’s attention for better patient outcomes. In the case of chronic kidney failure and automated PD, patients often require multiple, simultaneous prescriptions and periodic updates to those prescriptions. Thus, nephrologists routinely prescribe PD prescriptions to their patients and actively monitor the patients’ health progress and records to determine whether any updates or amendments to the patients’ PD prescriptions or treatments are warranted.

[0012] This is becoming an increasingly difficult task, however, as the number of kidney failure patients continues to increase while the availability of skilled professionals decreases. Furthermore, human generation of prescriptions is an error-prone process. In addition, when determining an appropriate prescription for a patient, a medical professional typically assesses the patient’ s body and considers the patient’ s medical reports and/or the patient’ s family medical history. But the prescription is determined for the patient using only that medical professional’s knowledge and expertise. Stated another way, a medical professional usually issues a prescription based on that medical professional’s training and on previous prescriptions he/she or his/her coworkers have issued to other patients, who may or may not be similar genetically or otherwise to the patient of interest. Moreover, medical professionals are often unable to immediately recognize or appreciate efficacies of prescriptions (particularly involving newer drugs or treatments) and/or the appropriateness of a particular prescription for a patient who shares several similarities to other patients on a regional or global scale. And medical professionals are often unable to consider several other factors (e.g., lifestyle habits, external or environmental conditions, etc.) that may affect the efficacy or appropriateness of a prescription because the medical professionals either do not have ready access to the information and/or because the medical professionals lack the processing power (in comparison to computers) to consider all of the information available in timely, efficient, and/or worthwhile manner. To address these concerns, the present technology collects and analyzes patient information from a variety of sources and inputs the data into an artificial intelligence (AI) and/or machine learning (ML) model. In some embodiments, the model is trained using data from thousands of other patients and/or previous prescriptions issued by thousands of other doctors. In turn, the model generates and outputs (e.g., without human intervention) one or more recommended prescriptions that are customized or tailored for a particular patient. The recommended prescriptions output from the model represent the model’s predictions for the most effective or successful prescriptions for that patient based, at least in part, on the patient information input into the model and/or on the similarity of the patient information to other information previously used to train the model. The present technology then solicits a medical professional (e.g., the patient’s doctor) for approval of the recommended prescription, and the medical professional is permitted to either approve the prescription as recommended by the model or update (e.g., modify) the recommended prescription. In some implementations, the present technology can use the medical professional’s feedback to further train the AI and/or ML model. The approved or updated prescription is then be issued to the patient.

[0013] Thus, not only can the present technology recommend prescriptions for a patient based, at least in part, on information related to that patient and his/her doctor’ s knowledge or experience, but the present technology can also recommend prescriptions for the patient based, at least in part, (i) on other patients who are similar to the patient and/or who may be located in other parts of the world and/or (ii) on the knowledge and expertise of the other patients’ doctors. In these and other implementations, the present technology can consider and process other sources of information (e.g., social media activity, financial data, news reports, lifestyle habits, genealogical information, and/or other data) that is either unavailable to the patient’s doctor or that the patient’s doctor is unable to fully process or appreciate for the purposes of generating and issuing prescriptions in a timely, efficient, and/or worthwhile manner. The present technology also (a) frees up time the medical professional would have spent assessing the patient, determining an appropriate prescription, generating the prescription, and/or researching new drugs or treatments for the patient, and (b) reduces the possibility that a prescription issued to a patient includes errors or is not appropriate for the patient. In other words, the present technology is expected to reduce an amount of time and effort typically required to generate, select, and/or manage prescriptions for a patient.

[0014] In addition to the above-described advantages, the present technology can additionally or alternatively analyze patient information gathered by one or more components of the system (e.g., in compliance with patient consent and/or data privacy laws) to identify patient mental health, lifestyle, physical health, and/or other concerns. For example, the present technology can track whether a patient completes, misses, or aborts a therapy session; analyze information or measurements collected during a therapy session to detect or identify infections or other health conditions; analyze patient behavior (e.g., physical activity levels, social media activity, etc.); and/or recommend potential lifestyle (e.g., activity, diet, drinking, etc.) changes. In some implementations, the present technology can generate and transmit reminders or notifications to the patient, the patient’s caregiver, the patient’s doctor, the patient’s family member, and/or another individual. For example, the present technology can remind the patient to perform a therapy session or take medication, the present technology can notify the patient’s doctor if the patient misses one or more therapy sessions, and/or the present technology can notify the patient and/or the patient’ s doctor if peritonitis or another infection is detected.

B. Selected Implementations of Artificial Intelligence and/or Machine 1, earning Systems.

Devices and Methods for Designing Patient Prescriptions

[0015] FIG. 1 is a partially schematic block diagram illustrating hardware and/or software components of a prescription software system 100 configured in accordance with various implementations of the present technology. As shown, the system 100 includes a patient PD cycler 102, a patient device 104, a doctor device 106, and a software cloud 110. The system 100 can further include a happiness index software module 112 and a patient genealogical software cloud or module 114. In these and other implementations, the system 100 can include one or more other hardware and/or software components (e.g., another patient medical device, such as a patient hemodialysis machine) in addition to or in lieu of one or more of the hardware and/or software components (e.g., the patent PD cycler 102) illustrated in FIG. 1.

[0016] The patient PD cycler 102 performs therapy (e.g., peritoneal dialysis) on patients. In some implementations, the patient PD cycler 102 can collect (e.g., measure) or receive (e.g., from the patient, from a caregiver or operator, from the software cloud 110, and/or from other components of the system 100) patient data or information. In these and other implementations, the patient PD cycler 102 can include a QR or barcode scanner and/or other input components (e.g., hardware buttons or input options, software buttons or input options, etc.) for receiving patient and/or health-related data. The patient PD cycler 102 can also send data to the software cloud 110 and/or generate a request for the software cloud 110 to generate a new prescription. In some implementations, the patient PD cycler 102 includes a communication interface that allows the patient PD cycler 102 to communicate with the software cloud 110 via one or more wired or wireless communication means (e.g., WiFi, broadband, etc.). A request for a new prescription generated by the patient PD cycler 102 can include a unique identifying code, patient identifying information, patient therapy information, and other information about the patient and/or the therapy. In some implementations, information included in the request can be encrypted by the patient PD cycler 102 before being transmitted to the software cloud 110.

[0017] The patient device 104 can be, among other things, a smart, wearable, and/or web- enabled device. For example, the patient device 104 can be a smartwatch, a fitness tracker, a mobile device or another device that can detect biometric, activity, location, and/or other information about the patient, such as patient heart rate, blood oxygen levels, step counts, or the like. The patient device 104 can communicate the detected information and/or associated patient identification data to the software cloud 110 through one or more wireless communication means, such as Wi-Fi, Bluetooth, or the like. In some implementations, the patient device 104 can encrypt (a) the biometric and/or other information and/or (b) the associated patient identification data, before transmitting the data to the software cloud 110. In these and other implementations, the patient device 104 can generate and/or send a request for a new prescription in addition to or in lieu of the patient PD cycler 102.

[0018] The doctor device 106 can be any type of computing device, such as a laptop computer, a desktop computer, a tablet computer, a smart phone, a smart wearable, or the like. The doctor device 106 can be used by a doctor or physician to communicate with the software cloud 110 and to receive notifications about a patient’s health, different patient health trends observed by the software cloud 110, prescription information for the patient, newly generated prescriptions for the patient, or the like. The doctor device 106 can allow a doctor or physician to approve or update a generated prescription and remain apprised of the patient’s therapy trends as the patient undergoes various therapeutic procedures.

[0019] The software cloud 110 refers to a collection of functional hardware and/or software modules that are stored within and/or operate using, at least in part, a network of (e.g., remote) servers, databases, and/or mainframe computers. As shown in FIG. 1, the software cloud 110 receives data from the patient PD cycler 102, the patient device 104, the doctor device 106, the happiness index software module 112, the patient genealogical software module 114, and/or other data sources. In some implementations, the software cloud 110 performs data analytics on all or a portion the received data. Based at least in part on the received data and/or on data analytics performed by the software cloud 110, the software cloud 110 can design (e.g., generate) a prescription for the patient. For example, the software cloud 110 can design a prescription for the patient by feeding all or a subset of the received data and/or one or more results from data analytics performed by the software cloud 110 into one or more artificial intelligence and/or machine learning models. In turn, the artificial intelligence and/or machine learning model(s) can recommend one or more prescriptions for the patient. The software cloud 110 and individual components of the software cloud 110 are described in greater detail below in relation to FIG. 3.

[0020] The happiness index software module 112 can provide to the software cloud 110 information about patient welfare, attitude, opinions, and/or emotions. For example, the happiness index software module 112 (a) can monitor and/or analyze a patient’s social media; investments; local, regional, and/or global news; and/or other information sources associated with the patient, and (b) present all or a subset of this data to the software cloud 110. All or a subset of the data sent to the software cloud 110 by the happiness index software module 112 can be used by the software cloud 110 during data analysis and/or prescription generation to select a proper prescription for the patient and/or recommend potential lifestyle changes. The happiness index software module 112 is described in greater detail below in relation to FIG. 3.

[0021] The patient genealogical software cloud module 114 can provide information to the software cloud 110 about patient health history and patient genealogy results. For example, a patient or a patient’s relative may undergo a genealogical DNA test (e.g., to identify specific locations of the patient’s genome, verify ancestral genealogical relationships, determine ethnicity or national origin, etc.). Genealogical data (e.g., ethnicity information) can provide insight into specific patient traits, such as peritoneal cavity size or responses to certain medications or treatments. Thus, the patient genealogical software module 114 can transmit genealogical data of the patient and/or the patient’s family member to the software cloud 110. In turn, all or a subset of the data sent to the software cloud 110 by the patient genealogical software module 114 can be used during data analysis and prescription generation to select and/or optimize a prescription for the patient. [0022] FIG. 2 is a flow chart illustrating a method 200 for generating a new prescription for a patient in accordance with various implementations of the present technology. In some implementations, the method 200 is executed in response to a user undergoing a new therapy or treatment. In other implementations, the method 200 can be executed in response to a patient or doctor requesting a new prescription for a patient. All or a subset of one or more of the steps of the method 200 can be executed by various components of a prescription software system, such as the prescription software system 100 of FIG. 1. For example, all of a subset of one or more of the steps of the method 200 can be performed by a patient PD cycle, a patient device, a doctor device, a software cloud, a happiness index software module, and/or a patient genealogical software module. In these and other implementations, all of a subset of one or more of the steps of the method 200 can be performed by a user of the prescription software system, such as a patient, an operator, a caregiver, a doctor, and/or a medical professional. Furthermore, any one or more of the steps of the method 200 can be executed in accordance with the discussion above and/or in accordance with the discussion below in relation to FIG. 3.

[0023] At block 201, the method 200 begins by activating a therapeutic machine. For example, a patient or operator can power on a patient PD cycler. In some implementations, after the therapeutic machine is activated, the therapeutic machine can establish a communication connection with a software cloud of a prescription software system.

[0024] At block 202, the method 200 continues by entering, collecting, and/or retrieving one or more patient data points or items. In some implementations, the one or more patient data points can be entered into, or collected or retrieved by, the therapeutic machine or patient device. For example, a patient or operator can enter the patient’s height, weight, age, body temperature, or the like into the therapeutic machine or patient device. Other examples of patient data points include patient identifying information, such as the patient’s name, gender, unique identifier, particular biometric data, or the like. The patient data points can additionally or alternatively include patient therapy information, such as ultrafiltration volume, number of PD or other therapeutic exchanges, dialysis solution concentration, fill time, fill volume, drain time, drain volume, therapy start time, therapy end time, and/or other dialysis or therapy information. In some implementations, patient data points can include external factors like weather, water quality, geographical location, local spice levels, humidity, room temperature, altitude, time of day, date, season, or the like. In these and other implementations, patient data points can include one or more patient health parameters, such as blood glucose, blood pressure, heart rate, water intake, lifestyle parameters, activity level (distance traveled, calories burned, steps taken, hours slept, stage of sleep reached, minutes active), ethnicity parameters, body mass index, travel history, comorbidities, health conditions, PET results, family medical history, and/or other parameters. Patient data points can be input into the system by the patient or operator; collected or measured (e.g., automatically or otherwise) by various instruments, such as the therapeutic machine, one or more patient smart, wearable, or web-enabled devices, and/or one or more medical devices; and/or retrieved from one or more databases or logs storing patient information. All or a subset of the data entered, collected, and/or retrieved at block 202 can be (automatically or at the direction of the patient/operator/caregiver) transmitted to a software cloud synchronously or asynchronously (as part of or independent of a new prescription request, which is described below in relation to blocks 203-205).

[0025] At block 203, the method 200 continues by requesting a new therapeutic prescription. In some implementations, a patient or operator can request a new therapeutic prescription by selecting a “Request New Prescription” hardware or software button option at the therapeutic machine and/or at the patient device. In these and other implementations, the therapeutic machine and/or the patient device can automatically generate and/or send a request for a new prescription. The therapeutic machine and/or the patient device can use all or a subset of the patient data points entered, collected, and/or retrieved at block 202 and/or other data points or items regarding the patient’ s current or most recent treatment to generate a new prescription request. The new prescription request can additionally or alternatively include an identifier of the therapeutic machine, an identifier of the patient’s most recent treatment, and/or other information for analysis.

[0026] At block 204, the method 200 continues by transmitting the generated prescription request to a software cloud via a wired or wireless communication connection. In some implementations, the therapeutic machine and/or the patient device transmit the new prescription request to the software cloud (e.g., directly or via one or more intermediary devices). In these and other implementations, the new prescription request is sent immediately after the new prescription request is generated at block 203. In other implementations, however, the new prescription request is not sent until after a user reviews and approves the request and initiates the transmission.

[0027] At block 205, the method 200 continues by generating a new prescription for the patient. In some implementations, the new prescription is generated based at least in part on the prescription request from blocks 203 and 204, on patient data points or information previously transmitted to the software cloud, and/or on information received from other components of the system. For example, an artificial intelligence and/or machine learning model can be used to generate the new prescription. Continuing with this example, the artificial intelligence and/or machine learning model can analyze the data included in the new prescription request of block 204 and/or other data input into the model, and can recommend a prescription for the patient. As noted previously, the machine learning model can consider the patient’s medical history, medication records, previous prescriptions or treatments, and/or other inputs, and can identify a prescription for the patient. In some implementations, the machine learning model can be trained using data from thousands or millions of training data items from other patients and doctors around the world with same or similar conditions or inputs requiring same or similar treatments and prescriptions. In some implementations, the model can output both (a) a recommended prescription and (b) one or more lifestyle changes and/or recommendations for the patient. For example, the model can output a recommended diet, amount of exercise (e.g., number of steps), hours of sleep, or the like in addition to a recommended therapy and/or medicinal prescription. Additional details regarding the generation of a prescription using an artificial intelligence and/or machine learning model are provided below in relation to FIG. 3.

[0028] At block 206, the method 200 continues by transmitting the generated/recommended prescription from block 205 to a doctor or another medical professional for review. The prescription can be sent to the doctor or the other medical professional using a wired or wireless communication connection between the software cloud and a computing device of the doctor and/or medical professional.

[0029] At decision block 207, the method 200 continues by determining whether the doctor and/or the other medical professional has approved the generated/recommended prescription from block 205. In some implementations, the doctor or medical professional reviews the recommended prescription on the doctor device 106 and makes a decision on whether or not to approve the prescription as recommended.

[0030] If the doctor or medical professional decides not to approve the prescription as recommended (“No” at block 207), the method 200 proceeds to block 208 where the doctor or medical professional can update (e.g., modify, alter, adjust, changes, etc.) the recommended prescription via user inputs. The updates can include changes to recommended dosage levels, changes to recommended medications, changes to recommended therapies, or the like. After the doctor or medical professional finishes updating the prescription, the method 200 can proceed to block 209 where the updated prescription can be transmitted back to the software cloud.

[0031] Referring again to block 207, if the doctor or medical professional approves the prescription as recommended by the software cloud (“Yes” at block 207), the medical professional can indicate his/her approval via user input, and the method 200 can proceed to block 209 where the approval can be transmitted to the software cloud without any changes to the recommended prescription.

[0032] At block 210, the method 200 continues by receiving and storing the approval and/or the updated prescription transmitted to the software cloud at block 209. In the event that an updated prescription is transmitted to the software cloud at block 209, the method 200 can update the recommended prescription for the patient at block 210. In some implementations, the artificial intelligence and/or machine learning model used to generate the recommended prescription at block 205 can be updated based at least in part on the user input received from the doctor or the medical professional at block 208. For example, if the doctor or medical professional provided updates to the prescription at block 208, the updates can be used to further train the artificial intelligence and/or machine learning model to improve future recommendations provided by the model. In another example, in the event the doctor or medical professional approves the recommended prescription without updates at block 207, the model can be further trained by reinforcing the prescription recommended by the model as a proper recommendation (e.g., by adjusting one or more weights assigned to neural layers of the artificial intelligence and/or machine learning models). The recommended prescription, the doctor’s or medical professional’s approval, and/or the updates to the recommend prescription provided by the doctor or medical professional can be stored in a database associated with the software cloud for later reference or use. For example, a doctor or medical professional can access the database at any time to review past, present, and/or future prescriptions generated for a patient. Continuing with the example, the doctor or medical professional can make updates to the prescriptions stored in the database at any time.

[0033] At block 211, the method 200 continues by transmitting the approved and/or updated prescription from block 210 to the patient. In some implementations, the prescription is sent to the patient’s therapeutic machine and/or to a patient device (e.g., a phone, a watch, a laptop, or other device). As discussed above, the prescription can specify a particular medication and dosage, a particular dialysis regimen, a particular therapeutic treatment, one or more lifestyle changes or recommendations, one or more physical fitness routines, or the like.

[0034] Although the steps of the method 200 are discussed and illustrated in a particular order, the method 200 of Figure 2 is not so limited. In other implementations, the steps of the method 200 can be performed in a different order. In these and other implementations, any of the steps of the method 200 can be performed before, during, and/or after any of the other steps of the method 200. Furthermore, a person skilled in the art will readily recognize that the method 200 can be altered and still remain within these and other implementations of the present technology. For example, one or more steps of the method 200 can be omitted and/or repeated in some implementations.

[0035] FIG. 3 is a partially schematic block diagram of a system 300 including a software cloud 310 (e.g., the software cloud 110 of FIG. 1) configured in accordance with various implementations of the present technology. As shown, the software cloud 310 can include a database 302, a machine learning module 304, a prescription generation module 306, a data analytics module 308, a medical charts module 309, and a data reception module 312.

[0036] The database 302 of the software cloud 310 can store data associated with generated prescriptions, patients, medical professionals, prescription requests, or the like. For example, the database 302 can receive, through the data reception module 312, data from a patient PD cycler 102 or therapeutic machine. As discussed above, the patient PD cycler 102 can send, to the software cloud 310, various patient data, such as patient weight, age, gender, body temperature, number of therapy exchanges, solution concentration, ultrafiltration volume, health records or parameters, and/or other data. In another example, the database 302 can store data from previous patient therapies and/or previous prescriptions issued to the patient from, for example, a doctor using a doctor device 106 (FIG. 1). The previous therapies and/or prescriptions can be stored in a previous patient prescriptions module 314. In these and other implementations, the database 302 can store data received from a happiness index software module 112, from a genealogical software cloud or module 114, from a patient device 104, or the like. In some implementations, the database 302 may also be accessed by the patient PD cycler 102, the doctor device 106, the genealogical software module or cloud 114, and/or other devices and modules, such as by a patient smart, wearable, and/or web-enabled device 104.

[0037] The machine learning module 304 of the software cloud 310 can utilize machine learning and/or other artificial intelligence to aid in generating prescriptions customized or tailored for a patient. The term “artificial intelligence” can be used interchangeably with “machine-learning” or “deep learning” in the context of this technology.

[0038] Machine learning systems are brain-inspired computing systems. A kidney failure patient on dialysis faces day to day challenges in the form of regular checkups and changes in prescription. Therefore, systems, devices, methods, and computer-readable mediums of the present technology are specifically constructed and/or programmed to suggest a customized prescription for each dialysis patient. Artificial intelligence is capable of learning similar to humans and can make decisions based upon that learning. In some cases, outcomes based on these artificial intelligence decisions turn out better than outcomes based on human decisions because artificial intelligence models (a) have access to a larger amount of data and/or (b) have the processing power (using one or more servers and/or mainframe computers) to analyze the large amount of data. For example, a doctor or medical professional typically assesses a patient’s body and considers his/her medical reports, family history, and/or other factors when determining an appropriate prescription for the patient. This approach, however, leaves room for improvement as the human brain cannot possibly consider every factor or source of information in timely, efficient, and worthwhile manner. In contrast, the machine learning module 304 of the present technology can consider not only a given patient's medical history and records (e.g., the patient’s past and/or present blood glucose, blood pressure, water intake, lifestyle, ethnicity, Body Mass Index (BMI), geographical condition and/or location, travel history, comorbidity, and/or other factors or data points), but also the past and/or present medical histories, records, prescriptions, and/or other information related to other patients on a global scale. Additionally, or alternatively, the machine learning module 304 can consider other sources of information, including non-medical information (e.g., the patient’s social media, investment accounts, local and global news, etc.) that a doctor or medical professional either cannot view or typically does not consider. Using the wealth of available information, the artificial intelligence models of the present technology can utilize different neural networks, decision trees, Support Vector Machines (SVMs), supervised and/or unsupervised algorithms, and/or clustering algorithms to learn to accurately classify, optimize, generate, and/or predict new prescriptions for patients. Prescriptions recommended by the artificial intelligence models can be individualized to the patient and/or can be session-specific or long-term prescriptions.

[0039] Customized prescriptions generated by the machine learning module 304 are used to help doctors optimize treatments for each patient. These optimizations are expected to improve overall patient outcomes in both the short and long terms. As discussed above, the machine learning module 304 in some implementations can additionally provide recommendations for improving a patient’s lifestyle and/or dietary consumption.

[0040] In detail, the machine learning module 304 of the software cloud 310 in FIG. 3 can learn from a vast amount of data provided to it by other components and/or modules of the software cloud 310 and/or by other components and/or modules of the system 300. With this data at its disposal, a neural network and/or one or more other algorithms of the machine learning module 304 can be trained by assigning weights to hidden neural layers based on defined parameters. In some implementations, the weights can be adjusted to reach desired outputs by testing the supervised neural network using several activation functions with backward and forward propagation. After the network is trained, validation data can be used to verify the machine learning module 304 is operating correctly. Once an acceptable validation acceptance accuracy percentage is met, the machine learning module 304 can then be used to generate patient prescriptions. The above learning process of the machine learning module 304 can continue as it continually collects data from a patient’ s PD cycler, therapeutic machine, and/or other sources, including user input provided by doctors or medical professionals related to prescriptions generated and/or recommended for the patient and/or for other (e.g., similar) patients around the globe and/or local to the patient of interest.

[0041] Outputs of the machine learning module 304 can be fed into the prescription generation module 306. The prescription generation module 306 can use one or more outputs of the machine learning module 304 to generate a prescription for a patient based at least in part on various input factors and/or the patient’s needs. For example, the prescription generation module 306 can optimize a prescription and specify various parameters of the prescription, such as a prescribed number of treatment cycles, a target ultrafiltration (“UF”) rate or volume, a dwell time, a drain time, fill volumes, and/or dextrose concentrations. The prescription generation module 306 can also allow doctors or medical professionals to interact with generated prescriptions, such as by permitting the doctors or medical professional to modify or update the generated or recommended prescriptions. In some implementations, the modifications and/or updates provided by doctors are used to update the machine learning module 304 (e.g., by using the modification and/or updates to adjust one or more of the weights of the neural network described above).

[0042] In some implementations, the prescription generation module 306 can generate prescriptions based on additional data. For example, the prescription generation module 306 can use patient happiness data, financial data, weather data, genetic data, geographic data, and/or other data. Patient happiness data can include data taken from patient social media activity, patient feedback, patient posting on social media, mental health data, eating data, drinking habits data, and/or current trends. Financial data can include stock market data, business performance data, and/or other data, and/or predictions for the how these financial data points affect the patient. Weather data can include weather forecasts and potential impact the weather can have on a patient’s mood. Patient genetic data can include data obtained from DNA testing of the patient that can indicate particular probabilities of health conditions and other genetic predispositions of the patient, including medical history of the patient and/or medical history of the patient’s family. Geographic condition and/or location data can include water quality data, altitude data, temperature data, humidity data, and/or other geographic parameters that can affect patient quality of life. Based on one or more of these factors, the prescription generation module 306 can recommend changes to a patient’s current prescription.

[0043] As discussed above, the prescription generation module 306 can send recommended prescriptions or a list of possible prescriptions to the patient’s doctor or another medical professional. In some implementations, the recommended prescription(s) sent by the prescription generation module 306 can include information related to suggested medications, including new medications the doctor or medical professional may not be familiar with.

[0044] The data analytics module 308 of the software cloud 310 can perform data analytics on all or a subset of the patient-related data for one patient or for a large group of patients, such as a group of patients associated with a doctor or medical professional or a group of doctors or medical professionals. The data analytics module 308 can track data related to therapies, treatments, prescriptions, and other analytics (e.g., patient happiness data, financial data, weather data, genetic data, geographic data, and/or other data) for patients. Various data analytics, such as determining a patient happiness index, can be performed based on this gathered data.

[0045] Dialysis patients commonly face many mental health challenges during their treatment, with depression and anxiety being the most commonly reported psychological conditions. Dialysis outcomes are often affected by patient psychological conditions and/or other external factors. For example, happy patients typically respond better and more quickly to dialysis treatment than stressed patients. One reason for this is that a patient’s mental condition can affect his/her eating or drinking habits, which in turn may negatively affect that patient’s responsiveness to dialysis treatment in comparison to other (e.g., happier) patients. The data analytics module 308 of the software cloud 310 of FIG. 3 can therefore calculate and/or monitor a patient’s happiness index values, which the software cloud 310 can use as data points or factors in predicting an appropriate prescription for the patient. In other words, different happiness index values can cause the software cloud 310 to generate, modify, and/or recommend different prescriptions for a patient. For example, the software cloud 310 (a) can recognize that a patient’s happiness index value is commonly high at a specific time each day in comparison to other times throughout the day, and (b) can recommend shifting the patient’s dialysis therapy schedule such that the dialysis therapy begins or ends at the specific time each day.

[0046] Each patient can have a patient profile stored within the data analytics module 308 and/or the database 302. The profile can contain various information related to therapy prescriptions, health records, biometrics, or the like. In some implementations, the patient can register or integrate their social media profiles to their patient profile. If given permission, the data analytics module 308 can use the patient’s public and/or private profile, posts, and/or activity as factors in determining the patient’s happiness index value. The data analytics module 308 can analyze the patient’s past social media activity and/or posting trends, and can actively compare it to the patient’s social media activity and/or posting trends since or while undergoing a therapy or treatment.

[0047] It is expected that an active social media user who routinely publishes positive posts and/or comments will continue to do so at a same or similar rate if he/she is satisfied with his/her therapy. On the other hand, it is expected that the patient will diverge from their usual social media activity if he/she is not satisfied with their therapy. For example, an agitated or unhappy patient is likely to publish negative posts and/or comments on a social media platform or might stop posting/commenting altogether. The data analytics module 308 can therefore monitor the patient’s social media activity to determine or identify a change in the patient’s happiness index value, and/or can recommend changes to the patient’s therapy (e.g., changing the patient’ s therapy start or end times) to attempt to change the patient’ s happiness index value back towards typical happiness index values for the patient. Additionally, or alternatively, the data analytics module 308 can alert the patient’s doctor, other medical professionals, and/or family members about the patient’ s mental state and/or behavior.

[0048] Social media activity on its own is often not enough to accurately determine a patient’s happiness index value. Thus, in some implementations, a patient’s happiness index value can additionally or alternatively be based on their locale, local news, global news, the stock market or the patient’s investment accounts, and/or other sources of information that are likely to affect the patient’s moods or stress levels. For example, in current times, coronavirus spread in an area local to the patient can affect the patient’s health or mental state. In addition to social media activity and news, the patient's economic wellbeing and employment status can be used as predictors of a patient’s mental health happiness index value.

[0049] In these and other implementations, the data analytics module 308 can analyze data it receives from various components of the system 300 and/or can send notifications or reminders to the patient, the patient’s doctor or medical professional, the patient’s caregiver, the patient’s family member, and/or other individuals. For example, a patient’s PD cycler 102 can track or log whether a patient completed, missed, or aborted one or more scheduled or prescribed dialysis therapy sessions. The PD cycler 102 can report this information to the software cloud 310, and the data analytics module 308 can generate and transmit (a) reminders to the patient or the patient’ s caregiver to perform or complete prescribed dialysis sessions and/or (b) notifications to the patient’s doctor or family member informing them of the information reported by the PD cycler 102. Additionally, or alternatively, the data analytics module 308 can track short-term and/or long-term health impacts of completing, missing, and/or aborting dialysis therapy sessions. The health impacts observed by the data analytics module 308 can be used to further inform or train the artificial intelligence model of the machine learning module 304. In these and other implementations, the data analytics module 308 analyze a patient’s session completion trends to (a) inform a happiness index value calculated for the patient and/or (b) identify potential mood, health, or other changes in the patient.

[0050] In some implementations, a patient’s PD cycler 102 can detect abnormalities or other causes for concern that occur during one or more dialysis sessions. For example, patients on peritoneal dialysis may contract different infections during therapy, with peritonitis being the most common and often caused by contamination of a disposable set or other therapy equipment. The PD cycler 102 can detect (e.g., using sensors or other devices/techniques) one or more conditions indicative of a patient infection. In response, the PD cycler 102 can report these conditions to the software cloud 310. In turn, the data analytics module 308 can analyze the reported conditions; determine the occurrence and/or identity of the infection; and/or notify the patient, the patient’s caregiver, the patient’s doctor or medical professional, the patient’s family member, and/or another individual. [0051] As another example, the patient PD cycler 102 can track various therapy data points over several sessions, which the data analytics module 308 can use to establish therapy trends for the patient and/or to identify deviations from those trends. As a specific example, the PD cycler 102 can track and report volumes of fluid (solution and toxins) drained from the patient during peritoneal dialysis sessions. The data analytics module 308 can determine an average range of volumes of fluid drained from the patient. If the reported volumes of fluid drained from the patient decrease over one or more dialysis therapy sessions, this can indicate a peritoneal cavity or other health condition. The data analytics module 308 can detect a deviation from the established drain volume trend for that patient, and can generate notifications (e.g., alerts) to the patient, the patient’s caregiver, the patient’s doctor or medical professional, the patient’s family member, and/or another individual. Additionally, or alternatively, the data analytics module 308 can (a) track the short-term or long-term health impacts of the decreased drain volumes on the patient, (b) provide the information to the machine learning module 304 (e.g., to further train the artificial intelligence model), and/or (c) provide the information to the prescription generation module 306 (e.g., for the prescription generation module 306 to determine whether to update the patient’s current or recommended prescription).

[0052] The medical charts module 309 of the software cloud 310 can use and/or quantify data analyzed by the data analytics module 308 to generate one or more medical charts pertaining to the patient. For example, the medical charts module 309 can generate one or more charts that track a patient’s happiness index value over time, a patient’s treatment over time, or a patient’s prescription over time. In these and other implementations, and the medical charts module 309 can output one or more of these charts for display to the patient, the patient’s caregiver, the patient’s doctor or another medical professional, and/or the patient’s family members to help track patient and/or treatment progress. In some implementations, the medical charts module 309 and/or the data analytics module 308 can remind patients, patient caregivers, and/or patient family members to perform particular treatments, perform particular therapies, and/or take prescribed medications.

[0053] The data reception module 312 of the software cloud 310 can perform data processing on various data received and/or utilized by the software cloud 310. For example, data can be cleaned and formatted, missing data can be acquired or estimated, data can be properly encoded, or the like. C. Examples

[0054] Several aspects of the present technology are set forth in the following examples. Although several aspects of the present technology are set forth in examples specifically directed to methods, computer-readable mediums, and systems; any of these aspects of the present technology can similarly be set forth in examples directed to any of systems, devices, methods, and/or computer-readable mediums in other implementations.

1. A method for generating a prescription using a machine learning model, the method comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and/or one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended prescription for the patient based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with the treatment of the patient; transmitting the recommended prescription to a medical professional for approval; and in response to receiving user input from the medical professional: generating a final prescription based on the received user input, and storing the final prescription for later transmittal to the therapeutic machine.

2. The method of example 1 wherein the one or more data items associated with the patient include blood glucose, blood pressure, water intake, lifestyle parameters, ethnicity parameters, body mass index, geographical condition, travel history, comorbidities, age, gender, height, weight, or any combination thereof.

3. The method of example 1 or example 2 wherein the one or more data items associated with a treatment of the patient include an ultrafiltration volume, a number of exchanges, a dialysis solution concentration, or any combination thereof.

4. The method any of examples 1-3 wherein: the machine learning model is trained using a plurality of training items; and each of the plurality of training items includes a prior patient, one or more data items associated with the prior patient, and a prescription given to the prior patient.

5. The method any of examples 1-4, further comprising updating the machine learning model based at least in part on the user input received from the medical professional.

6. The method of any of examples 1-5, further comprising making a modification to the final prescription based at least in part on a happiness index value associated with the patient.

7. The method of any of examples 1-6 wherein the final prescription includes a number of cycles, a target ultrafiltration value, a dwell time, a drain time, a fill volume, a dextrose concentration for a future treatment, or any combination thereof.

8. A non- transitory, computer-readable medium comprising instructions that, when executed by one or more processors, cause the one or more processors to execute a process for generating a prescription using a machine learning model, the process comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and/or one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended prescription for the patient based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with the treatment of the patient; transmitting the recommended prescription to a medical professional for approval; and in response to receiving user input from the medical professional: generating a final prescription based on the received user input, and storing the final prescription for later transmittal.

9. The non-transitory, computer-readable medium of example 8 wherein the one or more data items associated with the patient include blood glucose, blood pressure, water intake, lifestyle parameters, ethnicity parameters, body mass index, geographical condition, travel history, comorbidities, age, gender, height, weight, or any combination thereof. 10. The non-transitory, computer-readable medium of example 8 or example 9 wherein the one or more data items associated with a treatment of the patient include an ultrafiltration volume, a number of exchanges, a dialysis solution concentration, or any combination thereof.

11. The non-transitory, computer-readable medium of any of examples 8-10 wherein: the machine learning model is trained using a plurality of training items; and each of the plurality of training items includes a prior patient, one or more data items associated with the prior patient, and a prescription given to the prior patient.

12. The non-transitory, computer-readable medium of any of examples 8-11, the process further comprising updating the machine learning model based, at least in part, on the user input received from the medical professional.

13. The non-transitory, computer-readable medium of any of examples 8-12, the process further comprising making a modification to the final prescription based, at least in part, on a happiness index value associated with the patient.

14. The non-transitory, computer-readable medium of any of examples 8-13 wherein the final prescription includes a number of cycles, a target ultrafiltration value, a dwell time, a drain time, a fill volume, a dextrose concentration for a future treatment, or any combination thereof.

15. A computing system, the computing system comprising: one or more processors; and a memory, the memory comprising instructions that, when executed by the one or more processors, cause the one or more processors to execute a process for generating a prescription using a machine learning model, the process comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and/or one or more data items associated with a treatment of the patient, using the machine learning model to determine a recommended prescription for the patient based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with the treatment of the patient, transmitting the recommended prescription to a medical professional for approval, and in response to receiving user input from the medical professional: generating a final prescription based on the received user input; and storing the final prescription for later transmittal.

16. The computing system of example 15 wherein the one or more data items associated with the patient include blood glucose, blood pressure, water intake, lifestyle parameters, ethnicity parameters, body mass index, geographical condition, travel history, comorbidities, age, gender, height, weight, or any combination thereof.

17. The computing system of example 15 or example 16 wherein the one or more data items associated with a treatment of the patient include an ultrafiltration volume, a number of exchanges, a dialysis solution concentration, or any combination thereof.

18. The computing system of any of examples 15-17 wherein: the machine learning model is trained using a plurality of training items; and each of the plurality of training items includes a prior patient, one or more data items associated with the prior patient, and a prescription given to the prior patient.

19. The computing system of any of examples 15-18, the process further comprising updating the machine learning model based at least in part on the user input received from the medical professional.

20. The computing system of any of examples 15-19, the process further comprising making a modification to the final prescription based at least in part on a happiness index value associated with the patient. 21. The computing system of any of examples 15-20 wherein the final prescription includes a number of cycles, a target ultrafiltration value, a dwell time, a drain time, a fill volume, a dextrose concentration for a future treatment, or any combination thereof.

22. A method for generating a dialysate solution prescription using a machine learning model, the method comprising: receiving, from a therapeutic machine, a new dialysate prescription request for a patient, the new dialysate prescription request including one or more data items associated with the patient and/or one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended prescription for the patient based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with the treatment of the patient; and generating a final prescription specifying a sugar concentration for dialysate solution.

23. The method of example 22 wherein using the machine learning model to determine a recommended prescription includes determining the sugar concentration based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with a treatment of the patient.

24. The method of example 22 or example 23 wherein the final prescription includes instructions for adjusting an available sugar concentration to the sugar concentration.

25. The method of example 24 wherein the available sugar concentration is 1.5%, 2.3%, or 4.25%.

26. The method of any of examples 22-25 wherein the sugar concentration is not 1.5%, 2.3%, or 4.25%.

27. The method of any of examples 22-26 wherein the sugar concentration is a dextrose concentration. 28. The method of any of examples 22-27 wherein the dialysate solution prescription is a self-mixing dialysate solution prescription.

29. The method of any of examples 22-28, further comprising storing the final prescription for later transmittal to the therapeutic machine.

30. A method for generating a dialysate prescription with variable concentrations of electrolytes using a machine learning model, the method comprising: receiving, from a therapeutic machine, a new dialysate prescription request for a patient, the new prescription request including one or more data items associated with the patient and/or one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended prescription for the patient based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with the treatment of the patient; and generating a final prescription specifying an electrolyte concentration for dialysate.

31. The method of example 30 wherein the electrolyte concentration corresponds to a high calcium concentration.

32. The method of example 30 wherein the electrolyte concentration corresponds to a low calcium concentration.

33. The method of any of examples 29-32 wherein the dialysate prescription is a self mixing dialysate prescription.

34. The method of any of examples 29-33 wherein the electrolytes include calcium, sodium, and/or potassium.

35. The method of any of examples 29-34, further comprising storing the final prescription for later transmittal to the therapeutic machine. 36. A method for generating, using a machine learning model, a prescription specifying a dialysate solution concentration for one or more cycles of a dialysis treatment, the method comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and/or one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended dialysate solution prescription for the patient based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with the treatment of the patient; transmitting the recommended prescription to a medical professional for approval; and in response to receiving user input from the medical professional: generating a final prescription based on the received user input, and storing the final prescription for later transmittal to the therapeutic machine.

37. The method of example 36 wherein: the one or more cycles of the dialysis treatment include every cycle of the dialysis treatment; the recommended prescription and/or the final prescription include only one specific dialysate solution concentration or only one combination of specific dialysate solution concentrations; and the only one specific dialysate solution concentration or the only one combination of specific dialysate solution concentrations applies to each cycle of the one or more cycles.

38. The method of example 36 wherein: the one or more cycles includes a first cycle and a second cycle; the recommended prescription and/or the final prescription include: a first specific dialysate solution concentration or a first combination of specific dialysate solution concentrations for the first cycle, and a second specific dialysate solution concentration or a second combination of specific dialysate solution concentrations for the second cycle; and the first specific dialysate solution concentration or the first combination of specific dialysate solution concentrations is different from the second specific dialysate solution concentration or the second combination of specific dialysate solution concentrations, respectively.

39. A method for generating, using a machine learning model, a prescription for different patient behaviors, the method comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and/or one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended dialysate solution prescription for the patient based, at least in part, on the one or more data items associated with the patient, wherein the one or more data items associated with the patient include one or more data items related to one or more patient behaviors; transmitting the recommended prescription to a medical professional for approval; and in response to receiving user input from the medical professional: generating a final prescription based on the received user input, and storing the final prescription for later transmittal to the therapeutic machine.

40. A method for generating a prescription using a machine learning model, the method comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and/or one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended dialysate solution prescription for the patient based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with the treatment of the patient; transmitting the recommended prescription to a medical professional for approval; and in response to receiving user input from the medical professional: generating a final prescription based on the received user input, and storing the final prescription for later transmittal to the therapeutic machine, wherein the recommended prescription and/or the final prescription specify a mix of different solutions that are based, at least in part, on the one or more data items associated with the patient, the one or more data items associated with the treatment of the patient, and/or a prediction of patient outcome.

C. Conclusion

[0055] From the foregoing, it will be appreciated that specific implementations of the technology have been described herein for purposes of illustration, but well-known structures and functions have not been shown or described in detail to avoid unnecessarily obscuring the description of the implementations of the technology. To the extent any materials incorporated herein by reference conflict with the present disclosure, the present disclosure controls. Where the context permits, singular or plural terms can also include the plural or singular term, respectively. Moreover, unless the word “or” is expressly limited to mean only a single item exclusive from the other items in reference to a list of two or more items, then the use of “or” in such a list is to be interpreted as including (a) any single item in the list, (b) all of the items in the list, or (c) any combination of the items in the list. As used herein, the phrase “and/or” as in “A and/or B” refers to A alone, B alone, and both A and B. Where the context permits, singular or plural terms can also include the plural or singular term, respectively. Additionally, the terms “comprising,” “including,” “having” and “with” are used throughout to mean including at least the recited feature(s) such that any greater number of the same feature and/or additional types of other features are not precluded.

[0056] Furthermore, as used herein, the term “substantially” refers to the complete or nearly complete extent or degree of an action, characteristic, property, state, structure, item, or result. For example, an object that is “substantially” enclosed would mean that the object is either completely enclosed or nearly completely enclosed. The exact allowable degree of deviation from absolute completeness may in some cases depend on the specific context. However, generally speaking the nearness of completion will be so as to have the same overall result as if absolute and total completion were obtained. The use of “substantially” is equally applicable when used in a negative connotation to refer to the complete or near complete lack of an action, characteristic, property, state, structure, item, or result. Moreover, the terms “connect” and “couple” are used interchangeably herein and refer to both direct and indirect connections or couplings. For example, where the context permits, element A “connected” or “coupled” to element B can refer (i) to A directly “connected” or directly “coupled” to B and/or (ii) to A indirectly “connected” or indirectly “coupled” to B.

[0057] The above detailed descriptions of implementations of the technology are not intended to be exhaustive or to limit the technology to the precise form disclosed above. Although specific implementations of, and examples for, the technology are described above for illustrative purposes, various equivalent modifications are possible within the scope of the technology, as those skilled in the relevant art will recognize. For example, while steps are presented in a given order, alternative implementations can perform steps in a different order. As another example, various components of the technology can be further divided into subcomponents, and/or various components and/or functions of the technology can be combined and/or integrated. Furthermore, although advantages associated with certain implementations of the technology have been described in the context of those implementations, other implementations can also exhibit such advantages, and not all implementations need necessarily exhibit such advantages to fall within the scope of the technology.

[0058] It should also be noted that other implementations in addition to those disclosed herein are within the scope of the present technology. For example, implementations of the present technology can have different configurations, components, and/or procedures in addition to those shown or described herein. Moreover, a person of ordinary skill in the art will understand that these and other implementations can be without several of the configurations, components, and/or procedures shown or described herein without deviating from the present technology. Accordingly, the disclosure and associated technology can encompass other implementations not expressly shown or described herein.

Claims

CLAIMS What is claimed is:

1. A method for generating a prescription using a machine learning model, the method comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended prescription for the patient based, at least in part, on the one or more data items associated with the patient and the one or more data items associated with the treatment of the patient; transmitting the recommended prescription to a medical professional for approval; and in response to receiving user input from the medical professional: generating a final prescription based on the received user input, and storing the final prescription for later transmittal to the therapeutic machine.

2. The method of claim 1 wherein the one or more data items associated with the patient include blood glucose, blood pressure, water intake, lifestyle parameters, ethnicity parameters, body mass index, geographical condition, travel history, comorbidities, age, gender, height, weight, or any combination thereof.

3. The method of claim 1 wherein the one or more data items associated with a treatment of the patient include an ultrafiltration volume, a number of exchanges, a dialysis solution concentration, or any combination thereof.

4. The method of claim 1 wherein: the machine learning model is trained using a plurality of training items; and each of the plurality of training items includes a prior patient, one or more data items associated with the prior patient, and a prescription given to the prior patient.

5. The method of claim 1, further comprising updating the machine learning model based at least in part on the user input received from the medical professional.

6. The method of claim 1, further comprising making a modification to the final prescription based at least in part on a happiness index value associated with the patient.

7. The method of claim 1 wherein the final prescription includes a number of cycles, a target ultrafiltration value, a dwell time, a drain time, a fill volume, a dextrose concentration for a future treatment, or any combination thereof.

8. A non- transitory, computer-readable medium comprising instructions that, when executed by one or more processors, cause the one or more processors to execute a process for generating a prescription using a machine learning model, the process comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended prescription for the patient based, at least in part, on the one or more data items associated with the patient and the one or more data items associated with the treatment of the patient; transmitting the recommended prescription to a medical professional for approval; and in response to receiving user input from the medical professional: generating a final prescription based on the received user input, and storing the final prescription for later transmittal.

9. The non-transitory, computer-readable medium of claim 8 wherein the one or more data items associated with the patient include blood glucose, blood pressure, water intake, lifestyle parameters, ethnicity parameters, body mass index, geographical condition, travel history, comorbidities, age, gender, height, weight, or any combination thereof.

10. The non-transitory, computer-readable medium of claim 8 wherein the one or more data items associated with a treatment of the patient include an ultrafiltration volume, a number of exchanges, a dialysis solution concentration, or any combination thereof.

11. The non-transitory, computer-readable medium of claim 8-10 wherein: the machine learning model is trained using a plurality of training items; and each of the plurality of training items includes a prior patient, one or more data items associated with the prior patient, and a prescription given to the prior patient.

12. The non-transitory, computer-readable medium of claim 8, the process further comprising updating the machine learning model based, at least in part, on the user input received from the medical professional.

13. The non-transitory, computer-readable medium of claim 8, the process further comprising making a modification to the final prescription based, at least in part, on a happiness index value associated with the patient.

14. The non-transitory, computer-readable medium of claim 8 wherein the final prescription includes a number of cycles, a target ultrafiltration value, a dwell time, a drain time, a fill volume, a dextrose concentration for a future treatment, or any combination thereof.

15. A computing system, the computing system comprising: one or more processors; and a memory, the memory comprising instructions that, when executed by the one or more processors, cause the one or more processors to execute a process for generating a prescription using a machine learning model, the process comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and one or more data items associated with a treatment of the patient, using the machine learning model to determine a recommended prescription for the patient based, at least in part, on the one or more data items associated with the patient and the one or more data items associated with the treatment of the patient, transmitting the recommended prescription to a medical professional for approval, and in response to receiving user input from the medical professional: generating a final prescription based on the received user input; and storing the final prescription for later transmittal.

16. The computing system of claim 15 wherein the one or more data items associated with the patient include blood glucose, blood pressure, water intake, lifestyle parameters, ethnicity parameters, body mass index, geographical condition, travel history, comorbidities, age, gender, height, weight, or any combination thereof.

17. The computing system of claim 15 wherein the one or more data items associated with a treatment of the patient include an ultrafiltration volume, a number of exchanges, a dialysis solution concentration, or any combination thereof.

18. The computing system of claim 15 wherein: the machine learning model is trained using a plurality of training items; and each of the plurality of training items includes a prior patient, one or more data items associated with the prior patient, and a prescription given to the prior patient.

19. The computing system of claim 15, the process further comprising updating the machine learning model based at least in part on the user input received from the medical professional.

20. The computing system of claim 15, the process further comprising making a modification to the final prescription based at least in part on a happiness index value associated with the patient.

21. The computing system of claim 15 wherein the final prescription includes a number of cycles, a target ultrafiltration value, a dwell time, a drain time, a fill volume, a dextrose concentration for a future treatment, or any combination thereof.

23. The method of claim 22 wherein using the machine learning model to determine a recommended prescription includes determining the sugar concentration based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with a treatment of the patient.

24. The method of claim 22 wherein the final prescription includes instructions for adjusting an available sugar concentration to the sugar concentration.

25. The method of claim 24 wherein the available sugar concentration is 1.5%, 2.3%, or 4.25%.

26. The method of claim 22 where the sugar concentration is not 1.5%, 2.3%, or

4.25%.

27. The method of claim 22 wherein the sugar concentration is a specific dextrose concentration.

28. The method of claim 22 wherein the dialysate solution prescription is a self mixing dialysate solution prescription.

29. The method of claim 22, further comprising storing the final prescription for later transmittal to the therapeutic machine.

31. The method of claim 30 wherein the electrolyte concentration corresponds to a high calcium concentration.

32. The method of claim 30 wherein the electrolyte concentration corresponds to a low calcium concentration.

33. The method of claim 30 wherein the dialysate prescription is a self-mixing dialysate prescription.

34. The method of claim 30 wherein the electrolytes include calcium, sodium, and/or potassium.

35. The method of claim 30, further comprising storing the final prescription for later transmittal to the therapeutic machine.

36. A method for generating, using a machine learning model, a prescription specifying a dialysate solution concentration for one or more cycles of a dialysis treatment, the method comprising: receiving, from a therapeutic machine, a new prescription request for a patient, the new prescription request including one or more data items associated with the patient and/or one or more data items associated with a treatment of the patient; using the machine learning model to determine a recommended dialysate solution prescription for the patient based, at least in part, on the one or more data items associated with the patient and/or the one or more data items associated with the treatment of the patient; transmitting the recommended prescription to a medical professional for approval; and in response to receiving user input from the medical professional: generating a final prescription based on the received user input, and storing the final prescription for later transmittal to the therapeutic machine.

37. The method of claim 36 wherein: the one or more cycles of the dialysis treatment include every cycle of the dialysis treatment; the recommended prescription and/or the final prescription include only one specific dialysate solution concentration or only one combination of specific dialysate solution concentrations; and the only one specific dialysate solution concentration or the only one combination of specific dialysate solution concentrations applies to each cycle of the one or more cycles.

38. The method of claim 36 wherein: the one or more cycles includes a first cycle and a second cycle; the recommended prescription and/or the final prescription include: a first specific dialysate solution concentration or a first combination of specific dialysate solution concentrations for the first cycle, and a second specific dialysate solution concentration or a second combination of specific dialysate solution concentrations for the second cycle; and the first specific dialysate solution concentration or the first combination of specific dialysate solution concentrations is different from the second specific dialysate solution concentration or the second combination of specific dialysate solution concentrations, respectively.